This page documents production updates to Gemini Enterprise Agent Platform. Check this page for announcements about new or updated features, bug fixes, known issues, and deprecated functionality. You can see the latest product updates for all of Google Cloud on the [Google Cloud](https://docs.cloud.google.com/release-notes.md.txt) page, browse and filter all release notes in the [Google Cloud console](https://console.cloud.google.com/release-notes), or programmatically access release notes in [BigQuery](https://console.cloud.google.com/bigquery?p=bigquery-public-data&d=google_cloud_release_notes&t=release_notes&page=table). ## November 12, 2026 Feature **Gemini 4 Hydrogen (GA)** [Gemini 4 Hydrogen](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/4-hydrogen.md.txt) (`gemini-4-hydrogen`) is generally available (GA) in Gemini Enterprise Agent Platform as a short-term release (subject to a minimum 45-day retirement notice policy). Gemini 4 Hydrogen is built for speed and efficiency across everyday coding and agentic workflows, multimodal analysis, and tool use: - **Unified 1,048,576-token limit and long decodes** : Supports a unified 1,048,576-token input and output limit (where output includes thinking, function calls, and the final response). Single foreground segments generate up to 65,536 output tokens (streaming or non-streaming) and can be extended up to 1,048,576 tokens (minus input tokens) using long decodes (`continuation_token`) on both the [Interactions API](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/capabilities/interactions.md.txt) and the `GenerateContent` API (`v1beta1`), including asynchronous server-managed background execution (`background=true`) on the Interactions API. - **Thinking levels (`LOW`, `MEDIUM`, `HIGH`, and `XHIGH`)** : Supports four [thinking levels](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/thinking.md.txt), with `HIGH` as the default and `XHIGH` for deep reasoning workloads. Setting `thinking_budget` or `thinking_level="MINIMAL"` returns an HTTP `400` validation error. - **Implicit context caching with durable caching** : Supports automatic [implicit context caching](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/context-cache/context-cache-overview.md.txt) with durable caching enabled by default for shared prompt prefixes of 4,096 tokens or more (explicit context caching isn't supported). - **Cache write control and pricing** : Starting on January 12, 2027, Gemini 4 Hydrogen and later models support cache control features with TTL configuration. You choose what to cache and how long cached content stays valid through TTL configuration. Cache writes are billed at 1.25x the standard input token price by default, and longer retention costs more. For details, see [Pricing](https://cloud.google.com/model-garden/pricing#gemini-4). - **Office and OpenDocument file inputs** : Supports Microsoft Office (`.docx`, `.dotx`, `.doc`, `.pptx`, `.ppt`, `.xlsx`, `.xls`) and OpenDocument (`.odt`, `.odp`, `.ods`) files inline or from Cloud Storage (`gs://`), including inside multimodal `FunctionResponse` parts. - **Capabilities, consumption options, and endpoints** : Supports the `GenerateContent` API, the Interactions API, the OpenAI compatibility Chat Completions API, function calling, code execution, computer use, URL context, Grounding with Google Search, Google Maps, Enterprise Web Grounding, Parallel and Exa Web Search, Agentic video understanding, Standard PayGo with Usage Tiers, Priority PayGo, Flex PayGo, batch inference, and [Provisioned Throughput](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/supported-models.md.txt) (685 tokens per second per GSU) on the `global` endpoint and in the `us` and `eu` multi-regions. For implementation rules, long decode examples, and migration guidance from Gemini 3.8 Flash, see the [developer's guide to Gemini 4 Hydrogen](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/guides/gemini-4-hydrogen.md.txt) and [Get started with Gemini 4](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/start/get-started-with-gemini-4.md.txt). ## November 11, 2026 Feature **End-user safety identifier labels for abuse monitoring** End-user `safety_identifier` request labels are available in [General Availability (GA)](https://cloud.google.com/products#product-launch-stages) for Gemini (`generateContent` and `streamGenerateContent`) and Veo (`predictLongRunning`) API calls. API resellers and multi-tenant apps can pass an anonymized per-user ID in `labels` to attribute policy violations to individual end users. Targeted attribution helps prevent enforcement actions across an entire Google Cloud project. For formatting requirements and code samples, see [Pass an end-user safety identifier in API calls](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/abuse-monitoring.md.txt#pass-safety-identifier). Fixed **CodeMender updates (v0.21.0)** This release introduces updates to CodeMender: - **Sandboxed build commands and patch refresh** : When sandboxing is enabled, `cm build` and `cm fix` build validation now run `build.command` inside the sandbox, and `cm fix` refreshes stored patch diffs after build validation finishes. - **Read-only host toolchains and isolated caches** : Sandboxed `cm build`, `cm fix`, and `cm verify` runs now mount host toolchain and package-manager directories read-only and redirect supported compiler and package caches to session-scoped temporary directories. - **Command environment isolation** : Disabled repository Git hooks, isolated host Git configuration, and ignored repository Python bytecode caches (`__pycache__`) during command and build runs. - **Bug fixes** : - Added a 30-minute timeout for silent active turns so stalled sessions cancel and recover cleanly. - Preserved proof-of-concept execution output when reproducer errors occur, retaining stdout and stderr to help diagnose build failures, timeouts, and setup issues. - Fixed an issue where patch generation failed on mid-run sandbox failures by automatically retrying with a fresh sandbox environment. For more information, see the [CodeMender documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt). Feature **Gemini 4 Carbon in CodeMender (Preview)** CodeMender now supports Gemini 4 Carbon. To use it, pass `--model gemini-4-carbon` to `cm find`, `cm verify`, or `cm fix`. Gemini 3.8 Flash remains the default model. For more information, see [Specifying the model](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt#specifying-the-model). Feature **Govern agents on any CNCF-conformant Kubernetes cluster with Agent Gateway for cross-cloud (Limited Preview)** [Agent Gateway for cross-cloud](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/governance/agent-gateway-cross-cloud/overview.md.txt) is available in [Limited Preview](https://cloud.google.com/products#product-launch-stages). It extends Agent Platform governance to agents that run on Cloud Native Computing Foundation (CNCF)-conformant Kubernetes clusters outside of Google Cloud, such as Amazon Elastic Kubernetes Service (Amazon EKS), Azure Kubernetes Service (AKS), Red Hat OpenShift, and bare-metal clusters. Your agents keep running where they are, but you govern them from one place: give each agent a verifiable identity, register agents and MCP servers in Agent Registry, and route their LLM, MCP, and A2A traffic through local Agent Gateway proxies by using your existing service mesh or network stack. Those proxies enforce access and Model Armor safety policies, and send access logs and agent traces to Cloud Logging and Cloud Trace. Agent Gateway for cross-cloud is available at no charge during Limited Preview. Other products that you configure with it, such as Model Armor, Cloud Logging, and Cloud Trace, are billed at their standard rates. To request access, complete the [access request form](https://docs.google.com/forms/d/e/1FAIpQLSeUpnNRtYQjLRblii-KoUbJJ7DjEV6fvQAfO21IAXgVSqFjpw/viewform). For more information, see [Set up Agent Gateway for cross-cloud](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/governance/agent-gateway-cross-cloud/setup.md.txt). ## November 10, 2026 Feature **Cross-project governance for semantic governance policies** You can use semantic governance policies in a multi-project (hub-and-spoke) architecture to centrally govern agents deployed across multiple Google Cloud projects. When you configure a Resource Manager management boundary with the `AGENT_MANAGEMENT` capability, Agent Registry in the central management project automatically discovers agents across all member projects with their runtime identities intact. You can then provision a single Agent Gateway and semantic governance policy engine in the management project to author and enforce policies across member project agents. For more information, see [Govern agents across multiple projects](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/policies/configure-semantic-governance.md.txt#cross-project-governance) and [Cross-project boundaries](https://docs.cloud.google.com/agent-registry/boundaries-and-topology.md.txt). Feature **Provisioned Throughput supports more detailed order statuses** Provisioned Throughput supports more granular order statuses in the Google Cloud console in [General Availability (GA)](https://cloud.google.com/products#product-launch-stages). You can track order lifecycle states across `Declined`, `Cancelled`, and `Terminated`, in addition to `Pending review`, `Approved`, `Active`, and `Expired`. For more information, see [Check order status](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/purchase-provisioned-throughput.md.txt#check-order-status). Fixed **CodeMender updates (v0.20.0)** This release introduces updates to CodeMender: - **Bug fixes** : - Fixed patch application failures during validation on large repositories and projects with mixed LF and CRLF line endings. - Fixed scan error handling to properly surface permission denied errors when scanning unreadable directories instead of reporting zero findings. For more information, see the [CodeMender documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt). Feature **Gemini 4 Carbon and Gemini 4 Carbon for cyber (GA)** [Gemini 4 Carbon](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/4-carbon.md.txt) (`gemini-4-carbon`) and [Gemini 4 Carbon for cyber](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/4-carbon.md.txt#cyber) (`gemini-4-carbon-cyber-permissive`) are [generally available (GA)](https://cloud.google.com/products#product-launch-stages). Gemini 4 Carbon is Google's frontier model for complex, real-world tasks across coding, finance, legal, cybersecurity, and science, with a unified 1,048,576-token input and output limit, long decodes up to 1,048,576 tokens using `continuation_token`, configurable thinking levels (`LOW`, `MEDIUM`, and `HIGH`), implicit context caching with durable caching, Microsoft Office and OpenDocument file inputs, and [Provisioned Throughput](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/supported-models.md.txt#google-models) support. Gemini 4 Carbon for cyber is a defensive cybersecurity variant available behind the [Fairwind Program](https://deepmind.google/fairwind-program/) allowlist. ## November 09, 2026 Feature **Gemini 3.8 Flash regional support in Canada and Japan** [Gemini 3.8 Flash](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-8-flash.md.txt) (`gemini-3.8-flash`) is [generally available (GA)](https://cloud.google.com/products#product-launch-stages) in Canada (`northamerica-northeast1`) and Japan (`asia-northeast1`). On Gemini Enterprise Agent Platform, both regions support regional endpoints, in-region ML processing for data residency, and Provisioned Throughput. For more information, see [Deployments and endpoints](https://docs.cloud.google.com/gemini-enterprise-agent-platform/resources/locations) and [Data residency](https://docs.cloud.google.com/gemini-enterprise-agent-platform/resources/data-residency). Feature **Fast mode for Anthropic Claude models is generally available** [Fast mode](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude/fast-mode.md.txt) for supported Anthropic Claude models, starting with [Claude Fable 5.2](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude/fable-5-2.md.txt) (`claude-fable-5-2`), is available in [General Availability (GA)](https://cloud.google.com/products#product-launch-stages). You can set `"speed": "fast"` in online pay-as-you-go prediction requests to accelerate output token generation for latency-sensitive agentic workflows. For more information, see [Fast mode with Anthropic Claude models](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude/fast-mode.md.txt). Feature **Anthropic's Claude Fable 5.2** [Claude Fable 5.2](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude/fable-5-2.md.txt) (`claude-fable-5-2`) is available in [General Availability (GA)](https://cloud.google.com/products#product-launch-stages) in Model Garden, including support for [fast mode](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude/fast-mode.md.txt) (`"speed": "fast"`). Feature **Meta Muse models in Preview** [Muse Wonder from Meta](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/meta/muse-wonder-1-0.md.txt) (`meta/muse-wonder-1.0`), [Muse Ember from Meta](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/meta/muse-ember-1-0.md.txt) (`meta/muse-ember-1.0`), and [Muse Spark 1.3 from Meta](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/meta/muse-spark-1-3.md.txt) (`meta/muse-spark-1.3`) are available in [Preview](https://cloud.google.com/products#product-launch-stages) on the global endpoint through the OpenAI-compatible Responses API. Muse Wonder and Muse Spark 1.3 accept text, image, and PDF inputs, and Muse Ember accepts text and image inputs. All three models support reasoning, function calling, structured output, and a 1M-token context window. For more information, see [Muse models from Meta](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/meta.md.txt). ## November 06, 2026 Feature **Dedicated Memory Bank functionality in the Google Cloud console** You can create, configure, and inspect Agent Platform Memory Bank instances from the dedicated **Memory Bank** page in the Google Cloud console, with updated memory management controls for deployed agents in Agent Runtime: - **Memory Bank list and creation** : View and manage standalone memory banks on the **Memory Bank** page (separate from deployed agents on the **Deployments** page), edit basic information or labels, delete instances, and click **Create memory bank** to create an instance with default memory settings and Google-managed or Cloud KMS encryption. - **Memories inspection and filtering** : On the **Memories** tab, switch between **Unstructured** and **Structured** views, filter memories by `key:value` scope pairs or scope chips, edit or delete unstructured memories, inspect structured memory profiles and underlying facts in the **Memory details** panel, and select **View revisions** to inspect revision history. - **Instance and schema configuration** : On the **Configuration** tab, view and edit **Details** , **Generation** (memory extraction toggle, perspective, and managed or custom topics), **Retrieval** (embedding model, configurable before you store the first memory), and **Retention** (expiration and revision history TTL), or click **Schemas** to add and edit profile schemas. - **Agent integration snippets** : Click **Link to an agent** to get installation commands and prefilled Python code (populated with your project, region, and memory bank ID) for connecting with Agent Development Kit (ADK) or retrieving memories. - **Agent Runtime memory updates** : On the **Deployments** page, deployed agents include `key:value` scope filtering and **View revisions** on the **Memories** tab, along with **Generation** , **Retrieval** , and **Retention** configuration controls under **Service configuration** in the **Memory bank** section. For more information, see [Set up Memory Bank](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/memory-bank/setup), [Fetch memories](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/memory-bank/fetch-memories), [Memory Profiles](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/memory-bank/profiles), and [Memory revisions](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/memory-bank/revisions). ## November 05, 2026 Feature **Gemini Omni 1.2 Flash** [Gemini Omni 1.2 Flash](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/omni-1-2-flash.md.txt) (`gemini-omni-1.2-flash`) is available in [General Availability (GA)](https://cloud.google.com/products#product-launch-stages). Gemini Omni 1.2 Flash is an omni-modal foundation model that processes text, image, and video inputs to generate text and high-definition video with synchronized audio through the [Interactions API](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/capabilities/interactions.md.txt). Gemini Omni 1.2 Flash introduces the following capabilities: - **Standalone video upscaling** : Upscale previously generated videos from 360p to 720p or 1080p, and from 720p to 1080p or 4K using `task: "upscale"`. See [Upscale videos](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/video/upscale-videos.md.txt). - **Extended video generation and editing duration**: Generate or edit videos up to 15 seconds in length across text-to-video, image-to-video, reference-to-video, and video editing tasks (up to 10 seconds for first-and-last-frame interpolation and video extension). - **Reference-guided video editing and extension** : Supply up to 2 reference images along with an input video when editing or extending videos. See [Edit videos](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/video/edit-videos.md.txt) and [Extend videos](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/video/extend-videos.md.txt). - **Deterministic generation and billing labels** : Specify `seed` for reproducible video outputs and attach custom `labels` to Interactions API requests. - **Standard PayGo and Provisioned Throughput** : Available with [Standard Pay-as-you-go (PayGo)](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/standard-paygo.md.txt) and [Provisioned Throughput](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/gemini-omni-models.md.txt) (685 tokens per second per GSU). Feature **Interactions API long decodes for Gemini 4 Argon** Long decodes up to 1,048,576 tokens (minus input tokens) on the [Interactions API](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/capabilities/interactions.md.txt) (`v1beta1`) are available in [General Availability (GA)](https://cloud.google.com/products#product-launch-stages) for [Gemini 4 Argon](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/4-argon.md.txt) (`gemini-4-argon` and `gemini-4-argon-cyber-permissive`). You can extend single-turn outputs beyond the per-segment limit (65,536 streaming or 32,768 non-streaming tokens) using automatic continuation in the Google Gen AI SDK (`google-genai >= 2.34.0` for Python, `@google/genai >= 2.32.0` for JavaScript/TypeScript, Go `>= 1.77.0`, and Java `>= 1.80.0`), server-managed background execution (`background=true`), or custom stored (`store=true`) and stateless (`store=false`) `continuation_token` loops. For code samples, continuation options, and implementation rules, see [Long decodes in the developer's guide to Gemini 4 Argon](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/guides/gemini-4-argon.md.txt#long-decodes). Fixed **CodeMender updates (v0.18.0)** This release introduces updates to CodeMender: - **Verification tool-turn guardrails** : Added exploratory tool-turn limits to `cm verify` sessions to prevent unbounded command execution loops. - **Release notes link in update prompts**: Added a direct link to DevSite release notes when CodeMender prompts to update to a newer CLI version. - **Agent IAM diagnostic guidance** : Added actionable `roles/aiplatform.agentsUser` IAM binding commands to CLI setup and error messages for service account and CI/CD environments. - **Bug fixes** : - Improved the reliability of agent sessions, including verification and resumed sessions. - Improved compatibility across shell environments. - Improved the sign-in flow during interactive activation. - Improved error reporting when the service is temporarily unavailable. - Includes security and stability improvements. For more information, see the [CodeMender documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt). ## November 04, 2026 Feature **Provisioned Throughput support for supervised fine-tuned models (GA)** Provisioned Throughput [support for supervised fine-tuned models](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/supported-models.md.txt#supervised-fine-tuned-model-support) is [generally available](https://cloud.google.com/products#product-launch-stages). Burndown rates for fine-tuned model inference can include a premium that's directly proportional to the model's fine-tuned inference pricing premium. Feature **AlphaGenome gRPC and unified HTTP serving support** AlphaGenome (`alphagenome-004`) now supports unified HTTP/2 gRPC streaming and chunked binary protobuf HTTP streaming on Gemini Enterprise Agent Platform: - Streams compressed binary tensor chunks directly over gRPC on shared endpoints (`PredictionService.StreamDirectRawPredict` on port `8081`) and chunked binary protobuf payloads (`proto_b64`) over HTTP on dedicated endpoints (`streamRawPredict` on port `8080`) across all seven `DnaModelService` methods (`predict_sequence`, `predict_interval`, `predict_variant`, `score_variant`, `score_interval`, `score_ism_variants`, and `output_metadata`) and 11 genomic modalities, reducing latency for 1 MB track predictions. - Supports concurrent multi-variant evaluation for in silico saturation mutagenesis (`score_ism_variants`) without HTTP gateway timeouts. - Supports model fold selection (`ALL_FOLDS`, `FOLD_0`, `FOLD_1`, `FOLD_2`, and `FOLD_3`) across both gRPC and HTTP clients through the unified `alphagenome_cloud_sdk` Python SDK. - Retains `alphagenome-003` in Model Garden for backward compatibility with existing legacy HTTP deployments. For more information, see the [AlphaGenome documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/open-models/alphagenome.md.txt). ## November 02, 2026 Feature **Serving tuned Gemini 3.x models on global endpoints** Serving tuned Gemini 3.x models on `global` endpoints is available in [General Availability (GA)](https://cloud.google.com/products#product-launch-stages). You can copy fine-tuned Gemini 3.x models to the `global` region using the Model Registry `models:copy` API, deploy them to global serving endpoints, and send inference requests directly to `aiplatform.googleapis.com`. For more information, see [About supervised fine-tuning for Gemini models](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/tuning/supervised-tuning.md.txt) and [Tune Gemini models with supervised fine-tuning](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/tuning/supervised-tuning/use.md.txt#copy-tuned-model). ## October 30, 2026 Feature **Inference request metadata logging (GA)** Inference request metadata logging is generally available. Gemini Enterprise Agent Platform can write one Cloud Logging entry for every inference request your project serves. Each entry records the model, region, consumption option, token counts, Provisioned Throughput usage, latency and error code. Entries record metadata only and never contain prompts or responses. Logging is supported for Gemini models that use `generateContent` and `streamGenerateContent`, and for Claude models that use `rawPredict` and `streamRawPredict`. You can turn on logging in the Google Cloud console, from the Provisioned Throughput page, or with the REST API. Turning on logging in the console also turns on detailed metrics, which chart Provisioned Throughput traffic at one-second resolution. With logging on, the model observability dashboard adds 1-second resolution options to its charts for pay-as-you-go traffic. To attribute usage to a team, application or tenant, add request headers with the `x-goog-agent-platform-observability-label-` prefix, and query the entries with SQL in Observability Analytics. For more information, see [Log inference request metadata](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/capabilities/inference-metadata-logging). Feature **Computer use tool is available in General Availability** The [Computer use tool](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/tools/computer-use) is available in [General Availability (GA)](https://cloud.google.com/products#product-launch-stages). This release introduces comprehensive Human-in-the-Loop (HITL) safety governance, enabling developers to selectively configure or override specific confirmation policies (such as financial transactions or data modification) while allowing administrators to mandate organization-wide guardrails using the `configurableComputerUseToolSafetyPolicies` constraint. ## October 28, 2026 Change **Use a single regional Agent Registry instance for Gemini Enterprise apps** For Gemini Enterprise apps, we recommend associating each Agent Gateway with a single regional Agent Registry instance in the same project and region as the gateway. Use the `us-central1` registry for `global` and `us` apps, and the `europe-west1` registry for `eu` apps. This note updates the [September 9, 2026](https://docs.cloud.google.com/gemini-enterprise-agent-platform/release-notes#September_09_2026) entry. For more information, see [Register your agents and destination resources](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/gateways/set-up-agent-gateway#agent-registry). Fixed **CodeMender updates (v0.17.0)** This release introduces updates to CodeMender: - **Support bundle diagnostics** : Introduce `cm support-bundle` to collect environment metadata, configuration policies, health reports, and session logs into a compressed archive for streamlined troubleshooting. - **Plain output mode** : Add the persistent `--plain` flag (configurable via the `CM_PLAIN` environment variable or `output.plain` in `config.yaml`) to suppress decorative emojis, ANSI escape sequences, and Unicode borders for compatibility with screen readers and CI/CD log pipelines. - **Dedicated exit codes for setup failures**: Assign distinct exit codes (such as code 10 for credential issues) to configuration and authentication failures, preventing CI/CD pipelines from confusing environment setup issues with security findings. - **Tailored CI/CD credential guidance**: Provide environment-specific authentication guidance when running inside automated CI/CD systems such as GitHub Actions, GitLab CI, Buildkite, and Jenkins instead of interactive login prompts. - **Single-line JSONL statistics** : Add the `--jsonl` flag to `cm stats` to output single-line JSON records with model and team attribution for seamless integration with Cloud Logging and log aggregation pipelines. - **Granular sandbox and policy flags** : Replace the overloaded `--unrestricted` flag with orthogonal flags (`--sandbox=false`) and configurable command policies to decouple container sandboxing from execution safety controls. - **Sandbox configuration protections** : Mount `.git/hooks`, `.git/config`, and `.codemender` read-only within writable execution sandboxes during `cm fix` and `cm verify` to prevent modifications to repository security policies. - **Bug fixes** : - Added a security warning when commands are executed with both sandbox isolation disabled and interactive approval skipped. - Enforced `allowedRoots` path boundaries across host discovery and remediation file operations to prevent traversal via dangling symlinks. - Supplied exact finding line ranges to remediation agents during `cm fix` to eliminate manual file scanning overhead. - Updated `cm report import` to properly treat findings without location data as file-level findings rather than defaulting to the first line. - Ensured `cm session cancel` routes requests to the session's original region regardless of local environment overrides. - Fixed finding status tracking in `cm verify` so unexploitable verdicts update the local state database rather than leaving findings open. For more information, see the [CodeMender documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt). ## October 27, 2026 Feature **Gemini Robotics ER 2 (Preview)** [Gemini Robotics ER 2](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/gemini-robotics-er.md.txt) ([`gemini-robotics-er-2-preview`](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/robotics-er-2.md.txt)) is available in [Preview](https://cloud.google.com/products#product-launch-stages). Gemini Robotics ER 2 is an embodied reasoning vision-language model built on Gemini 3.5 Flash for robotics applications, including [spatial reasoning](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/gemini-robotics-er/spatial.md.txt), [agentic code execution](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/gemini-robotics-er/agentic.md.txt), [task orchestration](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/gemini-robotics-er/orchestration.md.txt), and [video understanding](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/gemini-robotics-er/video-understanding.md.txt). Feature **xAI's Grok 4.7 is generally available** [Grok 4.7](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/grok/grok-4-7.md.txt) is now [generally available (GA)](https://cloud.google.com/products#product-launch-stages) for production use on the global endpoint and the US multi-region endpoint. ## October 26, 2026 Feature **Control user intent matching for semantic governance policies** You can now turn user intent matching on or off for semantic governance policies by setting the `sgpUserIntentMatching` key on the authorization extension metadata. User intent matching is on by default; set the key to `DISABLED` to turn it off. Turning it off reduces protection against indirect prompt injection. When the key is set, policy evaluation logs report the mode each evaluation used in the `sgpUserIntentMatchingMode` field. For more information, see [Control user intent matching](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/policies/configure-semantic-governance.md.txt#control-intent-matching). Feature **Cloud Storage output support for Gemini image generation models** Direct Cloud Storage output persistence is [generally available (GA)](https://cloud.google.com/products#product-launch-stages) for [Gemini Nano Banana 2.1](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/nano-banana-2-1.md.txt) (`gemini-nano-banana-2.1`), [Gemini 3.1 Flash Image](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-1-flash-image.md.txt) (`gemini-3.1-flash-image`), [Gemini 3.1 Flash-Lite Image](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-1-flash-lite-image.md.txt) (`gemini-3.1-flash-lite-image`), and [Gemini 3 Pro Image](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-pro-image.md.txt) (`gemini-3-pro-image`). You can set `outputConfig.outputUriPrefix` and `generationConfig.responseFormat[].image.delivery` to `"URI"` in `generateContent` and `streamGenerateContent` requests to write generated and edited images directly to a Cloud Storage bucket and receive the object URI in `fileData.fileUri` instead of inline bytes. For more information, see [Save generated images to Cloud Storage](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/capabilities/image-generation.md.txt#save-generated-images-cloud-storage). Feature **Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS are Generally Available (GA)** [Gemini 3.8 Flash TTS](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-8-flash-tts.md.txt) and [Gemini 3.8 Flash-Lite TTS](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-8-flash-lite-tts) are available in [General Availability (GA)](https://cloud.google.com/products#product-launch-stages) in the `global` location and the `us` and `eu` multi-regions. The models support structured speaker and style metadata, inline vocal tags, multi-speaker dialogue, and streaming. [Voice design](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/text-to-speech/voice-design) and [voice replication](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/text-to-speech/voice-replication) capabilities remain available in Preview. For more information, see [Generate speech with Gemini TTS](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/text-to-speech/overview). Feature **Semantic governance policies support more model providers** Semantic governance policies now evaluate tool calls from agents that use models from providers other than Gemini. The policy engine reads each provider's request and response format, so the same policies apply whichever model your agent uses. Supported APIs include the Gemini Developer API and Gemini models on Gemini Enterprise Agent Platform, the Anthropic Messages API, Claude models on Gemini Enterprise Agent Platform, the OpenAI Chat Completions API, the Mistral AI API, the OpenAI-compatible endpoint on Gemini Enterprise Agent Platform (including Llama models), and self-hosted OpenAI-compatible servers such as vLLM. Streamed model responses aren't supported. To govern these providers, update the `when` condition of your authorization policy to match each provider's request path. The updated condition also matches requests that include a query string. For more information, see [Supported model providers](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/policies/semantic-governance-overview.md.txt#supported-providers) and [Connect the policy engine to Agent Gateway](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/policies/configure-semantic-governance.md.txt#connect-policy-engine). ## October 21, 2026 Fixed **CodeMender updates (v0.16.0)** This release introduces updates to CodeMender: - **Deep scan by default (`cm find`)** : `cm find` now runs a concurrent file-by-file deep scan (`--deep`) by default for higher vulnerability recall across repositories. To run a single-session quick scan instead, pass the `--quick` flag (`cm find --quick`). `cm stats` now breaks down scan usage by mode (`--deep`, `--quick`, and `--diff`). - **Pre-flight health checks (`cm health`)** : Added the `cm health` command (with `--json`, `--strict`, and `--timeout` flags) to verify credentials, configuration, endpoint reachability, and project authorization before running scans. - **Expanded file discovery** : `cm find` now scans `.hh`, `.hxx`, `.inl`, and `.ipp` C++ headers, extensionless scripts with supported language shebangs (`python`, `node`, `ruby`, and `php`), nested `build/`, `target/`, `obj/`, and `bin/` source directories, and production code under `spec/`, `locale/`, and `i18n/`. - **Bug fixes** : - Restricted `cm verify` file writes to `.exploit/` inside the project root and skipped missing or out-of-scope session artifacts during workspace restore. - Capped single-turn file reads at 256 KiB with pagination hints, bounded unified diff generation on large files, and raised `run_command` default and maximum timeouts to 120 seconds and 300 seconds. - Fixed an issue where sandboxed sessions could not write tool call entries to `~/.codemender/audit.log`. - Updated multi-file `--deep` and `--diff` scans to report `INCOMPLETE` status when one or more file scans fail. For more information, see the [CodeMender documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt). ## October 19, 2026 Feature **Supervised fine-tuning for Gemini 3 models is available in General Availability** [Supervised fine-tuning](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/tuning/supervised-tuning.md.txt) for Gemini 3 models is available in [General Availability (GA)](https://cloud.google.com/products#product-launch-stages). Key updates include the following: - **Gemini 3.5 Flash Lite support** : Supervised fine-tuning is now supported for [Gemini 3.5 Flash Lite](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-5-flash-lite.md.txt) (`gemini-3.5-flash-lite`), in addition to Gemini 3.5 Flash (`gemini-3.5-flash`) and Gemini 3.1 Flash Lite (`gemini-3.1-flash-lite`). - **Supported regions** : Model tuning jobs and tuned model serving for Gemini 3 models are supported on the `us` and `eu` multi-region endpoints. - **Customer-managed encryption keys (CMEK)** : [Customer-managed encryption keys (CMEK)](https://docs.cloud.google.com/gemini-enterprise-agent-platform/machine-learning/general/cmek.md.txt) are supported for protecting tuning job resources and tuned models. For more information, see [About supervised fine-tuning for Gemini models](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/tuning/supervised-tuning.md.txt) and [Tune Gemini models by using supervised fine-tuning](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/tuning/supervised-tuning/use.md.txt). ## October 14, 2026 Feature **Priority PayGo support for Gemini Nano Banana 2.1** [Priority Pay-as-you-go (PayGo)](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/priority-paygo.md.txt) is available in [General Availability (GA)](https://cloud.google.com/products#product-launch-stages) for [Gemini Nano Banana 2.1](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/nano-banana-2-1.md.txt) (`gemini-nano-banana-2.1`) on the `global` endpoint. Fixed **CodeMender updates (v0.15.0)** This release introduces updates to CodeMender: - **Enhanced SARIF 2.1.0 reporting** : Updated `cm report` SARIF output with standardized rule descriptors, severity properties, and repository-relative artifact locations for improved GitHub Code Scanning compatibility. - **Model invocation metrics** : Added model invocation tracking to `cm stats --json`, providing total, per-session, and per-turn invocation counts. - **Bug fixes** : - Fixed an issue where token usage from completed turns was omitted from session logs during error or cancellation exit paths. For more information, see the [CodeMender documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/codemender). Feature **DeepSeek-V4.1-Flash is available in Public Preview** DeepSeek-V4.1-Flash from DeepSeek is available as a fully managed model (MaaS) in Model Garden. The model accepts text and image input, targets agentic and coding tasks, and supports a context window of up to 1M tokens. For more information, see [DeepSeek-V4.1-Flash](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/maas/deepseek/deepseek-v41-flash.md.txt). ## October 13, 2026 Feature **Gemini 4 Argon in CodeMender (Preview)** CodeMender now supports Gemini 4 Argon. To use it, pass `--model gemini-4-argon` to `cm find`, `cm verify`, or `cm fix`. Gemini 3.8 Flash remains the default model. For more information, see [Specifying the model](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt#specifying-the-model). Feature **Gemini 4 Argon and Gemini 4 Argon Cyber Permissive** [Gemini 4 Argon](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/4-argon.md.txt) (`gemini-4-argon`) is available in [General Availability (GA)](https://cloud.google.com/products#product-launch-stages). Gemini 4 Argon is Google's frontier model that excels at deep work in complex, real-world tasks across domains including coding, finance, legal, cybersecurity, and science. It supports a unified 1,048,576-token input and output limit, configurable thinking levels (`LOW`, `MEDIUM`, `HIGH`), [long decodes](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/guides/gemini-4-argon.md.txt#long-decodes) up to 1,048,576 tokens using continuation tokens on the `GenerateContent` API, and [agentic video understanding](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/capabilities/video-understanding.md.txt#agentic-video-processing) in GA. [Gemini 4 Argon Cyber Permissive](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/4-argon.md.txt#cyber) (`gemini-4-argon-cyber-permissive`), a version of Gemini 4 Argon for defensive cybersecurity use cases, is also available in GA behind the [Fairwind Program](https://deepmind.google/fairwind-program/) allowlist. - **Known issue (long decodes)** : Long decoding (`continuation_token`) on the `GenerateContent` API (`generateContent` and `streamGenerateContent`) is supported only in the `v1beta1` API (not `v1`). The Google Gen AI SDK uses `v1beta1` by default unless `api_version='v1'` is specified in `http_options`; for direct REST calls, use the `v1beta1` endpoint path. For more information, see [Get started with Gemini 4](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/start/get-started-with-gemini-4.md.txt) and the [developer's guide to Gemini 4 Argon](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/guides/gemini-4-argon.md.txt). ## October 12, 2026 Fixed **CodeMender updates (v0.14.0)** This release introduces updates to CodeMender: - **Better pull request scans** : `cm find --diff` now catches removed security controls and gives large files more time (configurable with `--diff-job-timeout`). - **Unified sessions** : `--deep` and `--diff` scans now appear as a single session in `cm session list`, `cm report`, and `cm stats`. - **Moved code in `cm verify`** : `cm verify` now finds code that moved since `cm find` and updates the finding's location. - **Guided activation** : `cm` now opens the activation page and waits for activation instead of exiting. - **New `cm find` flags** : `--deep-max-turns` limits exploratory steps per file in `--deep` scans, and `--enable-bash` allows shell commands during scans. - **Improved detection**: More distinct findings per file and more precise CWE classification for weak hashing and randomness. - **Bug fixes** : - Fixed scans of repositories reached through nested symbolic links. - Fixed findings for nonexistent files and over-merged findings. - Shell scripts created by `cm verify` are now executable. For more information, see the [CodeMender documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt). ## October 07, 2026 Feature **Anthropic's Claude Haiku 5.5** [Claude Haiku 5.5](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude/haiku-5-5.md.txt) is available in Model Garden. Fixed **CodeMender updates (v0.13.0)** This release introduces updates to CodeMender: - **Consistent finding deduplication** : Improved finding deduplication consistency across repeated `cm find`, `--deep`, `--diff`, and `--parallel` scans. - **Bug fixes** : - Fixed an issue where deduplicated findings could appear in `cm report --status DISMISSED`. - Improved local state reliability during concurrent `cm find --diff` and `cm find --deep` scans. - Updated streaming error handling so non-retryable errors fail immediately instead of retrying. - Ensured local session state and logs are saved cleanly when a session is interrupted (`Ctrl+C`). For more information, see the [CodeMender documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt). ## October 06, 2026 Feature **Gemini Nano Banana 2.1** [Gemini Nano Banana 2.1](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/nano-banana-2-1.md.txt) (`gemini-nano-banana-2.1`) is available in [General Availability (GA)](https://cloud.google.com/products#product-launch-stages). Gemini Nano Banana 2.1 is optimized for high-speed multimodal image generation and editing, offering improved visual quality, prompt adherence, and text rendering across `1K`, `2K`, and `4K` output resolutions. ## October 05, 2026 Fixed **CodeMender updates (v0.12.0)** This release introduces updates to CodeMender: - **Hidden file scanning** : Added the `--include-hidden` flag to `cm find` (or `include_hidden` under `scan` in `config.yaml`) to include hidden directories and files, such as `.github/` and `.config/`, in vulnerability scans. - **Larger file support** : Raised the default file discovery size limit (`max_file_size_kb`) from 500 KB to 2 MiB so `cm find` no longer skips large source files. Existing configurations that use the previous 500 KB default are upgraded automatically, and `--deep` and `--diff` scans can now read beyond the inline preview of files larger than 64 KB. - **Report output for every scan mode** : `cm find --output` now writes the scan report for standard, `--deep`, `--parallel`, and `--diff` scans. - **Scoped fixes** : `cm fix` now prevents the agent from modifying CI/CD pipelines, static analysis configurations, build definitions, and repository configuration files that are outside the scope of the vulnerability being fixed. - **Accurate session token usage** : `cm` now reports the total token usage across all turns of a session, instead of only the final turn, so token counts in `cm stats` and session exports match billed usage. - **Bug fixes** : - Fixed duplicate findings in `cm report` and `cm stats`, and stopped merging distinct vulnerability types reported on the same line. - Fixed `database is locked` errors during `cm find --deep` scans with multiple workers. - Reduced false positives by requiring a concrete, viable exploit before reporting a finding with high confidence. - Included sessions from `cm find --deep` and `cm find --diff` workers in `cm stats`. - Fixed `cm verify` resolving relative finding and exploit artifact paths against the current working directory instead of the project root. For more information, see the [CodeMender documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt). ## September 30, 2026 Feature **Cloud Trace integration with Agent Gateway (Preview)** Agent Gateway integrates with Cloud Trace in [Preview](https://cloud.google.com/products#product-launch-stages) to provide end-to-end request observability for agent workloads. Enabling Cloud Trace gives you visibility into how requests travel from your agents through the gateway and across Google Cloud services, tools, agents, and MCP servers. For more information, see [Use Cloud Trace](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/gateways/monitor-agent-gateway.md.txt#use-cloud-trace). Feature **xAI's Grok 4.7** [Grok 4.7](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/grok/grok-4-7.md.txt) is available in [Preview](https://cloud.google.com/products#product-launch-stages) in Model Garden. Fixed **CodeMender updates (v0.11.0)** This release introduces updates to CodeMender: - **Tiered location configuration** : Added support for configuring the service location using the `CM_LOCATION` environment variable or `location` in `config.yaml` (defaulting to `global`), replacing the `--location` command-line flag. - **CI gate severity validation** : Updated `cm find` to validate `--fail-on` severity values (`CRITICAL`, `HIGH`, `MEDIUM`, `LOW`, or `NONE`) up front instead of silently ignoring misspelled severities. - **Sandboxed VCS branching and hardening** : Enabled `vcs_branch` creation and candidate branch cleanup inside the default sandboxed worker for `--architecture` sessions, and hardened VCS template expansion against shell metacharacters and command injection. - **Bug fixes** : - Normalized vulnerability type casing and relative file paths during `cm find` and consensus runs to prevent duplicate findings. - Extended HTTP 429 retry backoff (`Retry-After` support and up to 60-second quota refill windows) and automatic stream reconnection to improve scan resilience under rate limiting. - Preserved standalone `cm find`, `cm verify`, and `cm fix` sessions on interrupt (`SIGINT`/`SIGTERM`) so interrupted runs can be resumed with `cm session resume`, while still canceling parallel consensus worker operations. For more information, see the [CodeMender documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/codemender). Feature The App Topology API that provides agent topologies in Gemini Enterprise Agent Platform is now [generally available](https://cloud.google.com/products#product-launch-stages) This launch introduces the following changes: - A new query builder for customizing suggested quick queries or creating your own query on the **Topologies** page. Custom queries let you explore more data related to your agents, such as identity, alerts, vulnerabilities, and underlying infrastructure. For details, see [View topologies for a project](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/topology/view-project-topology.md.txt). - A streamlined experience for the [single-agent topology](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/topology/view-agent-registry-topology.md.txt) in Agent Registry. A topology with single-hop traffic to and from the agent automatically loads on the page. ## September 29, 2026 Feature **Custom denial messages for semantic governance policies** You can now configure a fixed **Denial message** on a semantic governance policy to show end users when the policy denies a request, in place of the policy denial rationale. The message can be up to 1,000 characters and is configurable in the Google Cloud console on the policy create or edit page, or by setting `agentResponseCustomization.denialMessage` on the policy resource using the REST API. When a single request triggers denials from more than one policy, the policy engine combines the configured Denial messages, deduplicates them, and falls back to a generic message for firing policies that don't have one configured. For more information, see [Configure a semantic governance policy](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/policies/configure-semantic-governance.md.txt) and [Information exposure risk](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/policies/semantic-governance-overview.md.txt#natural-language-constraints-nlc). ## September 28, 2026 Feature **Anthropic's Claude Sonnet 5.5** [Claude Sonnet 5.5](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude/sonnet-5-5.md.txt) is available in Model Garden. Feature **Provisioned Throughput: Support for changing order scope and increasing term** Provisioned Throughput now directly supports from the self service console the ability to change the scope or increase the term of an order through a supersede and replace operation. For more information, see [Restructure an order](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/purchase-provisioned-throughput.md.txt#restructure-order). Feature **Gemini 3 models supported by the Interactions API in Preview** Gemini 3 models are supported by the Interactions API in [Preview](https://cloud.google.com/products#product-launch-stages) on Gemini Enterprise Agent Platform. For more information, see the [Interactions API overview](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/capabilities/interactions.md.txt). Feature **Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS (Preview)** [Gemini 3.8 Flash TTS](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-8-flash-tts.md.txt) and [Gemini 3.8 Flash-Lite TTS](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-8-flash-lite-tts) are available in [Preview](https://cloud.google.com/products#product-launch-stages) on the global endpoint. The models support structured speaker and style metadata, inline vocal tags, multi-speaker dialogue, streaming, [voice design](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/text-to-speech/voice-design), and [voice replication](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/text-to-speech/voice-replication). For more information, see [Generate speech with Gemini TTS](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/text-to-speech/overview). ## September 25, 2026 Feature **Semantic governance policies support VPC Service Controls (Preview)** Semantic governance policy and the Semantic governance policy engine support [VPC Service Controls](https://docs.cloud.google.com/vpc-service-controls/docs/overview.md.txt) in [Preview](https://cloud.google.com/products#product-launch-stages). You can include Semantic governance policy and the policy engine in your service perimeter to help mitigate the risk of data exfiltration when governing agent tool calls. For more information, see [VPC Service Controls with Gemini Enterprise Agent Platform](https://docs.cloud.google.com/gemini-enterprise-agent-platform/machine-learning/general/vpc-service-controls#add-restricted-services) and [Semantic governance policies overview](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/policies/semantic-governance-overview). ## September 24, 2026 Feature **Gemini 3.8 Live** Gemini 3.8 Live is generally available. This release introduces improvements in voice quality, model reliability, and agent orchestration. For more information, see the following: - [Gemini 3.8 Live](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-8-live.md.txt) - [Gemini Live API overview](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/live-api.md.txt) Feature **Muse Spark 1.3 from Meta** [Muse Spark 1.3 from Meta](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/meta/muse-spark-1-3.md.txt) is available in [Preview](https://cloud.google.com/products#product-launch-stages). Muse Spark 1.3 from Meta is a reasoning model trained for agentic workflows and competitive coding. It delivers higher first-attempt accuracy, reliable built-in tool calling with MCP support, and 1M-token long context for multi-step tasks. Fixed **CodeMender updates (v0.10.0)** This release introduces updates to CodeMender: - **Impact-aware PR delta scanning** : Added `--diff` and `--staged` flags to `cm find` to audit modified pull request hunks and 1-hop neighbor files while suppressing pre-existing vulnerabilities. - **SARIF report export and CI gating** : Added SARIF v2.1.0 export (`--format sarif`, `--output`) and severity-based CI gating (`--fail-on`) to `cm report` to integrate scan results into CI/CD pipelines and code scanning tools. - **Adaptive hybrid deep scanning** : Added `--deep` and `--deep-workers` flags to `cm find` to combine deterministic file pruning, parallel package-batched analysis, and post-triage verification for higher vulnerability recall and precision. - **Custom workspace directory** : Added support for the `CM_HOME` environment variable to override the default `~/.codemender` directory for configuration, local state (`state.db`), backups, and artifacts. - **Bug fixes** : - Required build and test validation (`build.command`) to succeed during `cm fix` before marking a patch as applied and a finding as fixed. - Stabilized vulnerability fingerprints across multiple scans and bound `cm verify` directly to the target finding ID to prevent duplicate findings. - Restored the configured session model automatically when reconnecting to an active session with `cm session resume`. - Updated `cm verify` to add `.exploit` to `.gitignore` before execution so cancelled verification sessions do not leave untracked proof-of-concept scripts. For more information, see the [CodeMender documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt). ## September 22, 2026 Feature **Anthropic's Claude Opus 5.5** [Claude Opus 5.5](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude/opus-5-5.md.txt) is available in Model Garden. ## September 21, 2026 Fixed **CodeMender updates (v0.9.0)** This release introduces updates to CodeMender: - **Interactive HTML security reports** : Overhauled `cm report --format html` to provide a modern, interactive dashboard featuring severity metric cards, syntax-highlighted code snippets with line numbers, and an inline patch diff viewer. Added the `--open` (`-o`) flag to automatically open the generated report in the default browser. - **Expanded language support** : Added out-of-the-box vulnerability scanning support for C# (`.cs`), Rust (`.rs`), Kotlin (`.kt`, `.kts`), Ruby (`.rb`), and PHP (`.php`) to the default discovery configuration and initialization templates. - **Per-turn latency metrics** : Enhanced `cm stats` and session exports to report per-turn latency breakdowns, distinguishing time spent waiting on model inference from local tool execution. - **Bug fixes** : - Improved session reliability and error recovery during long-running repository scans. - Fixed local workspace state compatibility issues when upgrading from earlier CLI versions. For more information, see the [CodeMender documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt). ## September 18, 2026 Feature **xAI's Grok 4.6 is generally available** [Grok 4.6](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/grok/grok-4-6.md.txt) is now [generally available (GA)](https://cloud.google.com/products#product-launch-stages) and available for production use on the global endpoint and the US multi-region endpoint. Breaking **Agent Platform SDK for Python version 2.0.1 is available** Version 2.0.1 of the Agent Platform SDK for Python (`google-cloud-agentplatform`) is now available. This release migrates generative AI modules to the Google Gen AI SDK, decouples the agent surface from `google-cloud-aiplatform` into a dedicated package, and introduces restructured namespaces. For details and migration instructions from `google-cloud-aiplatform`, see the [Agent Platform SDK for Python version 2.0.1 migration guide](https://docs.cloud.google.com/gemini-enterprise-agent-platform/machine-learning/python-sdk/sdk-migration.md.txt). ## September 17, 2026 Feature **Gemini Omni Flash supports stateful and streaming video generation (Preview)** Gemini Omni Flash supports stateful (`store: true`) and server-sent event (SSE) streaming (`stream: true`) video generation in the Interactions API in [Preview](https://cloud.google.com/products#product-launch-stages). You can temporarily store generated videos and interaction state on the server, stream status updates and final outputs over an SSE connection, or retrieve completed asynchronous interactions using unary or streaming GET requests. For more information, see [Generate videos from text](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/video/generate-videos-from-text). ## September 16, 2026 Feature **CodeMender updates (v0.8.0)** This release introduces updates to CodeMender: - **Gemini 3.8 Flash default** : Gemini 3.8 Flash (`gemini-3.8-flash`) is now supported and enabled as the default model for CodeMender CLI sessions, delivering faster inference and improved reasoning. A one-time notice in the CLI informs users when the new default is active. - **Tool payload guardrails**: Introduced safe output limits for file reading (2 MiB) and codebase grep search (512 KiB) with centered match context windows, eliminating payload overflow errors and improving stability during large repository scans. - **Bug fixes** : - Fixed an issue where shell detection and command execution on Windows could fail during repository resets and exploit verification when Git Bash was installed in standard registry or non-PATH locations. - Fixed an issue in `cm verify` where verified findings could report "not found" or fail to persist confidence and status upon session completion or resumption. - Fixed syntax errors in generated verification scripts caused by invalid regex escaping in grep assertions during `cm verify`. - Prevented HTTP 409 lease conflict errors during long-running sessions by ensuring streaming HTTP connections are promptly released. For more information, see [CodeMender documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt). Feature **Gemini 3.8 Flash Cyber available in allowlisted GA** Gemini 3.8 Flash Cyber is a post-training version of the Gemini 3.8 Flash model tailored specifically for cybersecurity use cases. Gemini 3.8 Flash Cyber is available in GA behind an allowlist. For more information, see [Gemini 3.8 Flash Cyber](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-8-flash-cyber). ## September 15, 2026 Feature **Reinforcement learning fine-tuning in the Google Cloud console (Preview)** You can create, monitor, and test reinforcement learning fine-tuning jobs for Gemini models in the Google Cloud console ([Preview](https://cloud.google.com/products#product-launch-stages)). From the **Models \> Tuning** page, you can configure Python code or model-based reward functions, test reward logic against sample prompts before launching a job, track training and evaluation metrics in real time, and test tuned checkpoints in Agent Studio. For more information, see [Quick start: Reinforcement learning fine-tuning using the console](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/tuning/reinforcement-tuning/quick-start-console.md.txt). ## September 14, 2026 Deprecated **Gemini model deprecation and retirement date updates** The retirement and deprecation dates for the following Gemini models have been updated: - **Gemini 2.5 Flash Image (`gemini-2.5-flash-image`)** : Deprecated and scheduled for retirement on March 15, 2027 (extended from October 2, 2026). Migrate to [Gemini 3.1 Flash-Lite Image](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-1-flash-lite-image.md.txt) (`gemini-3.1-flash-lite-image`). - **Gemini 3.1 Flash-Lite Image (`gemini-3.1-flash-lite-image`)**: Retirement date is scheduled for June 28, 2027 or later. For more information, see [Gemini model versions and lifecycle](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/model-versions.md.txt). Fixed **CodeMender updates (v0.7.0)** This release introduces updates to CodeMender: - **Network stream resilience**: Improved CLI session stability with automatic reconnection and transient error recovery during long-running scans and remediation workflows. - **Configuration uniformity** : Standardized directory exclusion rules across configuration files and CLI scanning flags under `scan_config.exclude_dirs`. - **Bug fixes** : - Fixed an issue where `cm report` incorrectly categorized `DISMISSED` findings as `OPEN` in the summary table. - Resolved sandbox permission denial errors by preventing child worker processes from attempting to create internal session logs on disk. - Hardened sandbox command policy to prevent directory traversal and file inspection outside the designated repository root into adjacent directories. For more information, see [CodeMender documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt). Feature **Cyber Verification Program for Claude is available in Preview** Anthropic's Cyber Verification Program (CVP) is available in [Preview](https://cloud.google.com/products#product-launch-stages) on Gemini Enterprise Agent Platform. CVP enables verified organizations to use supported Claude models (Claude Opus 4.7, Claude Opus 4.8, Claude Sonnet 5, and Claude Opus 5) for legitimate defensive cybersecurity tasks with default dual-use restrictions lifted. For more information, see [Cyber Verification Program for Claude](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude/cyber-verification-program). ## September 10, 2026 Feature **Standard PayGo adds Tier 4** Tier 4 is added to the Standard PayGo usage tiers, for spend above $50,000 per 30-day period, and Tier 3 is bounded at $2,000 to $50,000 per 30-day period. For details, see [Standard PayGo usage tiers](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/standard-paygo.md.txt#usage-tiers-and-throughput). Change **Priority PayGo removes ramp limits** Priority PayGo throughput limits are available immediately, with no ramp-up period, and traffic above the limit is downgraded to Standard PayGo only when there isn't spare capacity to serve it at priority. For details, see [Priority PayGo throughput limits](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/priority-paygo#throughput-limits). Feature **Provisioned Throughput: Support for multiple pending orders and change requests** Provisioned Throughput now directly supports from the self service console the ability to schedule multiple new orders at the same time and change requests across multiple orders within the same project. For more information, see [Purchase Provisioned Throughput](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/purchase-provisioned-throughput.md.txt). ## September 09, 2026 Feature **Computer Use and Shell sandboxes are generally available** [Computer Use](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/sandbox/computer-use.md.txt) and Shell sandboxes in Gemini Enterprise Agent Platform are now generally available (GA). This release also includes the following new features for Agent Platform sandboxes: - **Shell sandboxes** : Run untrusted shell commands, install packages, and manipulate files in an isolated Linux container using direct API `/exec` calls. For more information, see the [Shell sandbox quickstart](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/sandbox/shell-sandbox-quickstart.md.txt). - **VPC Service Controls \& Private Service Connect** : Protect sandbox data and isolate network boundaries with VPC Service Controls, private ingress endpoints (PSC-E), and private egress routing (PSC-I). For more information, see [Configure VPC Service Controls and Private Service Connect with sandboxes](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/sandbox/configure-vpc-sc.md.txt). - **Customer-Managed Encryption Keys (CMEK)** : Protect sandbox data at rest, including disk storage and snapshot checkpoints, using Cloud KMS keys. For more information, see [Configure customer-managed encryption keys (CMEK) for sandboxes](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/sandbox/configure-cmek.md.txt). - **Pausing and resuming sandboxes** : Deschedule compute resources for idle sandboxes while preserving file system state and connection identity, and resume them in seconds. For more information, see [Manage sandboxes](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/sandbox/manage-sandboxes.md.txt#pause-a-sandbox). Feature **Priority PayGo now supports the US and EU multi-region endpoints** You can send Priority PayGo requests to the `us` and `eu` multi-region endpoints, in addition to the `global` endpoint. For more information, see [Priority PayGo](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/priority-paygo.md.txt). Feature **Agent Gateway supports VPC Service Controls** Agent Gateway now enforces VPC Service Controls perimeter rules for agent communications. When you [configure Agent Gateway with VPC connectivity](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/gateways/set-up-vpc-connectivity), agent traffic is routed through your private VPC network, ensuring that your organization's VPC-SC perimeter rules are applied to all agent traffic as well. Note that setting up VPC connectivity is required to enable VPC Service Controls perimeter enforcement for Agent Gateway deployments. The connectivity template must be configured in `ALL_TRAFFIC` egress mode. > [!IMPORTANT] > **Important:** VPC Service Controls is only supported for Agent Gateway deployments created after September 8, 2026 that use the [agent connectivity > template](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/gateways/set-up-vpc-connectivity) to configure VPC connectivity. Feature **Agent Gateway supports multiple Agent Registry instances** Agent Gateway now lets you associate up to two Agent Registry instances (one global registry and one regional or multi-region registry) with a single Agent Gateway instance. For more information, see [Register your agents and destination resources](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/gateways/set-up-agent-gateway#agent-registry). > [!IMPORTANT] > **Important:** There has been an update to this feature. For the corrected note, see the entry for [October 28, > 2026](https://docs.cloud.google.com/gemini-enterprise-agent-platform/release-notes#October_28_2026). ## September 08, 2026 Feature **Agent connectivity templates for VPC connectivity in Agent Gateway** Agent Gateway now uses *agent connectivity templates* (`agentConnectivityTemplate`) to configure and manage egress connectivity to VPC networks. Connectivity templates let you choose whether to route all outbound agent traffic (`ALL_TRAFFIC`) or only traffic to certain private IP address ranges (`PRIVATE_RANGES_ONLY`) into your VPC network. For more information, see [Set up VPC connectivity for Agent Gateway](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/gateways/set-up-vpc-connectivity). ## September 02, 2026 Feature **Gemini 3.8 Flash is generally available** [Gemini 3.8 Flash](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-8-flash.md.txt) is now generally available (GA) and available for production use. For more information on 3.8 Flash, see the [model page](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-8-flash). Feature **Provisioned Throughput support for Gemini Omni** Provisioned Throughput supports Gemini Omni. To learn more, see [supported models](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/supported-models.md.txt). Fixed **CodeMender updates (v0.6.0)** This release introduces updates to CodeMender: - **Machine-readable metrics** : Added the `--json` flag to `cm stats` to export aggregate and per-session metrics (`CACHE_HIT%`, `THINK_RATIO%`, `TOOL_CALLS`, `DURATION`). - **Session drill-down** : Added `cm stats --session ` to inspect turn-by-turn token consumption for specific sessions. - **Bug fixes** : - Improved codebase search reliability by skipping binary archives and non-regular files during traversal. - Fixed an issue where `cm report import` failed on native JSON reports or findings referencing new files. - Fixed an issue where preview mode could create empty directories on disk before user confirmation. - Prevented erroneous verification verdicts when workspace reset fails. For more information, see [CodeMender documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt). Feature **Deferred tier for autonomous agent scheduling (Preview)** The *deferred tier* is available in Preview. The deferred tier automatically queues non-latency-sensitive background agent workloads and schedules them to execute during off-peak hours. Key capabilities and benefits include: - **50% token discount**: Receive a 50% discount on consumed tokens for deferred workloads. - **Reduced rate limiting**: Mitigate resource exhaustion (429) errors and infrastructure pressure during long-running background tasks. - **Supported agents** : - **Deep Research Agent** : Pass `service_tier="deferred"` in the Python SDK or `"service_tier": "deferred"` in REST API interaction requests. For more information, see [Autonomous agent scheduling](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/efficiency/autonomous-scheduling.md.txt). ## September 01, 2026 Feature **Anthropic's Claude Fable 5.1 is now available** [Claude Fable 5.1](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude/fable-5-1.md.txt) is available in Model Garden. Change **Embedding SKUs are changing** Memory bank uses the following embedding model SKUs: - 6E46-5623-C0B6 - C15D-A68F-5C1E - 2D07-D52C-A93B For detailed information about individual SKUs, see [Google Cloud Platform SKUs](https://cloud.google.com/skus). Feature **Agentic video processing now available in Preview** Gemini 3.5 Pro and later models now support agentic video processing, where the model dynamically navigates video content instead of extracting frames at a fixed rate. This reduces token usage for long-form video and improves answer quality. Set `processing: "agentic"` (Interactions API) or `media_processing: "AGENTIC"` (GenerateContent API) on video inputs, or use the model default. See [Video understanding](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/capabilities/video-understanding#agentic-video-processing) for details. Change **Pricing updates for agent metering** Session and memory bank compute metering is in effect for the Agent Platform compute SKU. For more information, see [Gemini Agent Platform Pricing](https://services.google.com/fh/files/emails/b_502770571_gemini_enterprise_agent_platform_pricing.pdf). ## August 31, 2026 Feature **Gemini 3.1 Flash Image and Gemini 3 Pro Image updates** This release introduces feature updates and expanded endpoint availability for Gemini 3.1 Flash Image and Gemini 3 Pro Image: - **Multi-region endpoint support** : [Gemini 3.1 Flash Image (`gemini-3.1-flash-image`)](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-1-flash-image.md.txt) is now available on US (`us`) and EU (`eu`) multi-region endpoints for model availability, ML processing, Provisioned Throughput, and PayGo Standard. - **4K image output in GA** : Generating 4K resolution images is now [Generally Available (GA)](https://cloud.google.com/products#product-launch-stages) for both [Gemini 3.1 Flash Image](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-1-flash-image.md.txt) and [Gemini 3 Pro Image](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-pro-image.md.txt). - **Video inputs in GA** : Image generation from video inputs is now [Generally Available (GA)](https://cloud.google.com/products#product-launch-stages) for [Gemini 3.1 Flash Image](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-1-flash-image.md.txt). For more information, see the following: - [Gemini 3.1 Flash Image](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-1-flash-image.md.txt) - [Gemini 3 Pro Image](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-pro-image.md.txt) - [Model locations](https://docs.cloud.google.com/gemini-enterprise-agent-platform/resources/locations.md.txt#multi-region) Feature **IAM Unified Access Policies are generally available (GA)** IAM Unified Access Policies (Access policies) for Gemini Enterprise Agent Platform are [generally available (GA)](https://cloud.google.com/products#product-launch-stages). Agent Gateway uses Identity-Aware Proxy and Access policies to govern agentic communication between agent principals and destination resources, such as Model Context Protocol (MCP) servers, other agents, and registered or unregistered endpoints. Key capabilities include: - Multiple allow and deny rules within a single Access policy to establish fine-grained behavioral guardrails. - Common Expression Language (CEL) conditions in rules to enforce access criteria based on tool names, read-only constraints, HTTP methods, and URL path attributes. - Dry-run and enforcement modes to validate and audit policy evaluation before blocking agent traffic. - End-to-end agent identity authentication and authorization using mutual TLS (mTLS) and Context-Aware Access (CAA) with Demonstrating Proof of Possession (DPoP). For more information, see [IAM access policies overview](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/policies/iam-overview-uap.md.txt). ## August 27, 2026 Feature **CodeMender updates (v0.5.0)** This release introduces updates to CodeMender: - **Model support** : Support for Gemini 3.6 Flash (`gemini-3.6-flash`) and Gemini 3.7 Flash (`gemini-3.7-flash`, default). - **Unrestricted verification** : Added the `--unrestricted` flag to `cm verify` to bypass command policy restrictions during exploit verification in isolated environments. - **Bug fixes** : - Improved reliability of long sessions. - Improved shell resolution across operating environments. - Clearer permission denied error messages. - Fixed an issue where the CodeMender sandbox would fail to initialize with an error due to relative paths. For more information, see [CodeMender documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt). Feature **Gemini Omni 1.1 Flash is available in Public Preview** [Gemini Omni 1.1 Flash (`gemini-omni-1.1-flash-preview`)](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/omni-1-1-flash.md.txt) is available in [Preview](https://cloud.google.com/products#product-launch-stages). Gemini Omni 1.1 Flash is a multimodal model designed for video, image, and text tasks, optimized for high-speed video generation that supports audio and video editing. ## August 24, 2026 Feature **GLM 5.2 is available in Public Preview** GLM 5.2 from Z.ai is available as a fully managed model (MaaS) in Model Garden. The model targets long-horizon agentic and coding tasks and supports a 1M-token context window. For more information, see [GLM 5.2](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/maas/zaiorg/glm-52). ## August 21, 2026 Feature **xAI's Grok 4.6** [Grok 4.6](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/grok/grok-4-6.md.txt) is available in [Preview](https://cloud.google.com/products#product-launch-stages) in Model Garden. ## August 18, 2026 Feature **CodeMender updates (v0.4.0): Model support** This release introduces updates to CodeMender: - **Gemini 3 Flash removal** : Gemini 3 Flash (`gemini-3-flash-preview`) is no longer supported as a model backend for CodeMender. CodeMender supports Gemini 3.5 Flash (default) and Gemini 3.1 Pro Preview. For more information, see [Specifying the model](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender.md.txt#specifying-the-model). ## August 15, 2026 Feature **Monitor semantic governance policies with built-in metrics (Preview)** Built-in Cloud Monitoring metrics for the semantic governance policy engine are available in Preview. You can now observe request throughput, evaluation counts, latencies, verdict distribution (`ALLOW` versus `DENY`), and LLM token consumption for the policy engine directly in [Metrics Explorer](https://docs.cloud.google.com/monitoring/charts/metrics-explorer.md.txt), query them through the Cloud Monitoring v3 API and PromQL, and use them in alerting policies. For more information, see [Monitor semantic governance policies](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/policies/monitor-semantic-governance). Feature **AlphaFold 3 is generally available** AlphaFold 3 is now generally available (GA) and available for approved users for commercial research use. AlphaFold 3 Google DeepMind's and Isomorphic Labs' deep learning model to predict the 3D structures and interactions of proteins, DNA, RNA, ligands, and ions. For more information, see [AlphaFold 3](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/open-models/alphafold-3.md.txt). ## August 13, 2026 Feature **Gemini 3.7 Flash is generally available** [Gemini 3.7 Flash](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-7-flash.md.txt) is now generally available (GA) and available for production use. For more information on 3.7 Flash, see the [model page](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-7-flash). Feature **Agent Runtime: ADK telemetry metrics** Agents deployed to Agent Runtime with Agent Development Kit (ADK) 2.6.0 or later can emit `gen_ai` application metrics that follow OpenTelemetry's generative AI semantic conventions. Set the `GOOGLE_CLOUD_AGENT_ENGINE_ENABLE_TELEMETRY` environment variable to export these metrics to Cloud Monitoring as user-defined metrics, alongside agent traces and logs. For more information, see [Set up monitoring](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/runtime/monitoring.md.txt#adk-metrics). ## August 12, 2026 Feature **CodeMender CLI: Sandbox enabled by default** This release updates the CodeMender CLI default behavior: - **Sandbox enabled by default** : The CLI now runs commands inside the process-level sandbox by default to protect your workstation. You can disable the sandbox in your `config.yaml`, by passing `--sandbox=false` to CLI commands, or bypass it using the `--unrestricted` flag. For more information, see [Install the CLI and configure](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender/set-up-environment.md.txt). ## August 04, 2026 Feature **Process-level sandboxing and auto-updates (Preview)** This release introduces process-level sandboxing, automatic update checks, and other improvements to the CodeMender CLI (Preview): - **Process-level sandboxing** : To safeguard your workstation against unintended file modifications or unexpected tool side effects, you can now execute all agent-proposed tools (such as compiling code, running tests, or executing shell scripts) inside an OS-level sandbox. Sandboxing is disabled by default to allow seamless dependency resolution, but can be enabled persistently in `config.yaml` or per-command using the new `--sandbox` flag. - **Automatic updates** : The CLI now automatically checks for updates in the background (at most once every 24 hours) when running in an interactive terminal. You can also run `cm update` to forcefully check for and apply updates immediately. - **Token usage statistics** : You can now use the `cm stats` subcommand to view a summary of token usage (input, output, cached, thought, tool-use) for your local sessions. - **Improved Windows support**: The standalone CLI now has improved support for running on Windows. For more information, see [Install the CLI and configure](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender/set-up-environment.md.txt). ## July 29, 2026 Feature **Gemini Enterprise: Feedback service (Preview)** The Feedback service on Gemini Enterprise Agent Platform is available in Preview. The Feedback service lets you collect, analyze, and manage end-user qualitative feedback, such as thumbs-up and thumbs-down sentiments and feedback labels from interactions between your agents and users. You can also view feedback entries alongside traces in the console or export them to Cloud Trace to troubleshoot your agents. For more information, see [Feedback service overview](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/feedback-service). ## July 24, 2026 Feature **Anthropic's Claude Opus 5** [Claude Opus 5](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude/opus-5.md.txt) is available in Model Garden. ## July 21, 2026 Feature **Gemini 3.6 Flash and 3.5 Flash-Lite are generally available (GA)** [Gemini 3.6 Flash](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-6-flash.md.txt) and [Gemini 3.5 Flash-Lite](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-5-flash-lite) are now generally available (GA) and available for production use. These models are designed to improve upon their predecessors' capabilities, including improved token usage and improved document understanding. See the linked model information pages for more information. This release includes some potentially breaking changes from previous Flash and Flash-Lite models: - **Sampling parameters**: Custom values for temperature, top-K, and top-P are not supported and will be ignored if set. - **Penalty parameters**: Custom values for frequency and presence penalty parameters are not supported. Setting these will result in an API error. - **API turn structure** : API requests where the last input turn has a role of `Model` are no longer supported and will return an error: - **Interactions API** : Requests where the last object in the input array has `"type": "model_output"` will fail. - **GenerateContent API** : Requests where the last object in the contents array has `"role": "model"` will fail. Deprecated **Open model endpoint deprecations** The following open model endpoints are deprecated and will be retired on October 21, 2026. For more information, see [Open model deprecations](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/deprecations/open-models). - `deepseek-ocr-maas` - `deepseek-r1-0528-maas` - `deepseek-v3.2-maas` - `deepseek-v3.1-maas` - `glm-5-maas` - `glm-4.7-maas` - `gpt-oss-20b-maas` - `kimi-k2-thinking-maas` - `llama-3.3-70b-instruct-maas` - `minimax-m2-maas` - `multilingual-e5-large-instruct-maas` - `multilingual-e5-small-maas` - `qwen3-235b-a22b-instruct-2507-maas` - `qwen3-coder-480b-a35b-instruct-maas` - `qwen3-next-80b-a3b-instruct-maas` - `qwen3-next-80b-a3b-thinking-maas` Feature **CodeMender available in Public Preview** CodeMender is now available in [Preview](https://cloud.google.com/products#product-launch-stages). CodeMender is an AI code security agent that can find, verify, and fix deep vulnerabilities in your codebase. CodeMender wraps a fine-tuned harness around a large language model (LLM), using Google DeepMind-engineered prompts, skills, and orchestration logic to turn the model into an agentic system specialized in code security. Key capabilities include: - **Find vulnerabilities**: Scan your codebase using an agent-guided LLM with specialized security tools, or import findings from external static analysis tools. - **Verify vulnerabilities**: Build code and execute proof-of-concept (PoC) exploits in your local sandbox to confirm exploitability and reduce false positives. - **Fix vulnerabilities**: Automatically generate, validate, and apply source code patches tested against local builds and unit tests to prevent regressions. - **Manage sessions and diffs**: Track scan and remediation attempts as stateful sessions, inspect unified diffs, and export detailed findings reports in HTML, Markdown, JSON, or SARIF formats. For more information, see [CodeMender overview](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender). Announcement **CodeMender is available in Public Preview** CodeMender is now available in Public Preview for select customers. CodeMender is an AI-powered code security agent that runs on your local workstation to find, verify, and fix security vulnerabilities in your codebase. Key capabilities in this release include: - **Scan and find**: Scan your codebase for application logic, memory management, and authentication vulnerabilities, or import findings from external scanners. - **Verify**: Build your code and run tests locally to confirm whether findings are exploitable, minimizing false positives. - **Fix**: Auto-generate security patches and verify them locally before applying them to your workspace. - **Safety controls**: Enforces a Human-in-the-Loop workflow with mandatory manual confirmations for all disk writes and command executions by default. For more information, see the [CodeMender overview](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/codemender). ## July 20, 2026 Security **Security update for Server-Side Request Forgery (SSRF) in Agent Studio** This release fixes a Server-Side Request Forgery (SSRF) vulnerability in the auto-generated `/api-proxy` backend endpoint for web applications created before July 1, 2026, using Agent Studio. If you downloaded, generated, or deployed web application code from Agent Studio before July 1, 2026, regenerate the app from Agent Studio and deploy the new version. For more information, see [Quickstart: Deploy your Agent Studio prompt as a web application](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agent-studio/deploy-vais-prompt) The updated backend code includes strict domain allowlist validation, ensuring that destination hostnames for the `/api-proxy` endpoint end with allowed Google Cloud domains, such as `*-aiplatform.clients6.google.com` ## July 17, 2026 Breaking **Gemini 3.1 Flash Image Preview and 3 Pro Image Preview are retired** The Nano Banana preview models `gemini-3.1-flash-image-preview` and `gemini-3-pro-image-preview` have been retired and are no longer accessible. Update your code to use either `gemini-3.1-flash-image` or `gemini-3-flash-image` instead. ## July 15, 2026 Feature **Memory Bank memory profiles are generally available (GA)** Memory profiles in Memory Bank are generally available (GA). Memory profiles allow you to generate structured profiles, which are data structures with static schemas populated and updated using LLMs. By defining a fixed schema, you ensure your agents have immediate, low-latency access to evolving information without the need for expensive search operations during a session. For details, see [Memory profiles](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/memory-bank/profiles). ## July 09, 2026 Feature **Retirement for preview models for 2.5 Flash, 2.5 Flash-Lite, and 3.1 Flash-Lite** The following preview model endpoints have been retired and are no longer accessible: - `gemini-2.5-flash-lite-preview-09-2025` - `gemini-2.5-flash-preview-05-2025` - `gemini-3.1-flash-lite-preview` See [Migrate to the latest Google models](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/migrate) for information on how to migrate your project. ## July 08, 2026 Deprecated **Grok 4.1 models deprecation** The Grok 4.1 model family (including `xai/grok-4.1-fast-reasoning` and `xai/grok-4.1-fast-non-reasoning`) on the Gemini Enterprise Agent Platform is deprecated and will be shut down on August 20, 2026. After this date, Google Agent Platform Model as a Service (MaaS) will no longer serve these models. Specifically, API requests directed to the `https://aiplatform.googleapis.com/v1/projects//locations/global/endpoints/openapi/chat/completions` endpoint using the Grok 4.1 model IDs will fail and return a `400` error. To maintain service, migrate your applications to newer xAI models (such as Grok 4.2 or Grok 4.3) or choose an alternative model from the Google Cloud Model Garden. Feature **Memory Bank IngestEvents is generally available (GA)** The Memory Bank `IngestEvents` API is generally available. The `IngestEvents` API decouples event ingestion from memory generation, letting you continuously stream content to Memory Bank and configure when memory generation is triggered. This GA release includes the following features: - **Carry over context between generation windows:** Use the `overlap_event_count` parameter to re-include already-processed events at the start of the next window to keep memories coherent. - **Configure memory revisions for ingested events:** Use `revision_labels`, `revision_ttl`, or `disable_memory_revisions` to customize how generated revisions are managed. - **Attach metadata to memories:** Use the `metadata` and `metadata_merge_strategy` configuration parameters to store structured information alongside generated memories. For details, see [Ingest events](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/memory-bank/ingest-events.md.txt). Feature **Memory Bank support for Gemini Embedding 2** Memory Bank supports the `gemini-embedding-2` model for similarity search configurations. When you configure `gemini-embedding-2` as the embedding model, you must use one of the `global`, `us`, or `eu` endpoints in the model's resource name (for example, `projects/{project}/locations/us/publishers/google/models/gemini-embedding-2`). Memory Bank does not support using regional locations (for example, `us-central1`) for `gemini-embedding-2`. For details, see [Similarity search configuration](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/memory-bank/setup#similarity-search-config). ## July 06, 2026 Feature **AlphaGenome released for Gemini Enterprise Agent Platform** AlphaGenome, Google DeepMind's state-of-the-art genomics foundation model, is now available for deployment and use with Gemini Enterprise Agent Platform. Designed to decipher the functional regulatory code of the human genome, AlphaGenome analyzes large-scale DNA sequences at single-base resolution to predict how genetic variations affect molecular and biological mechanisms like gene expression, chromatin accessibility, and RNA splicing. For information on how to use AlphaGenome in Agent Platform, see the [documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/open-models/alphagenome.md.txt). ## July 01, 2026 Feature **Provisioned Throughput: Multiple pending new orders GA** The ability to submit multiple pending orders is generally available. you can submit up to seven Google model orders for the same model and region. See [Multiple pending orders](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/purchase-provisioned-throughput#multiple). ## June 30, 2026 Feature **Anthropic's Claude Sonnet 5** [Claude Sonnet 5](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude/sonnet-5.md.txt) is available in Model Garden. ## June 29, 2026 Feature **Gemini 3.5 Flash default model for Memory Bank** The default model used for Memory Bank generation is now Gemini 3.5 Flash instead of Gemini 2.5 Flash. For information, see [Set up Memory Bank](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/memory-bank/setup.md.txt). Feature **Semantic Governance Policies available in Public Preview** Semantic governance policies and the policy engine are now available in [Preview](https://cloud.google.com/products#product-launch-stages). Semantic governance policy provides an intelligent security and compliance layer that evaluates an AI agent's proposed tool calls against user intent and organizational business rules at runtime. Key capabilities include: - **Natural Language Constraints (NLC):** Author declarative business rules and security guardrails in plain English without needing to write code or redeploy agent applications. - **Layered Intent Gating:** Intercepts agent tool calls at runtime to verify alignment with trusted user intent and prevent unauthorized actions, rogue tool use, and data exfiltration. - **Granular Scoping:** Apply constraints globally across all tools for an agent or target specific tools and parameters (e.g., enforcing strict financial limits or geographic restrictions). - **Agent Skills Lifecycle Governance:** Protects agents from context poisoning and supply-chain exploits by governing the dynamic loading of Agent Skills (tool packages) during sessions. - **Dry Run Mode:** Test and observe policy verdicts in Log Explorer before enforcing them on active traffic. For more information, see [Semantic governance policies overview](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/policies/semantic-governance-overview.md.txt). Feature **Provisioned Throughput: Email notifications GA** The ability to get email notifications for Provisioned Throughput events is generally available. See [Get email notifications](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/use-provisioned-throughput#email-notifications). ## June 26, 2026 Feature **New Provisioned Throughput features** The following Provisioned Throughput features are generally available: - **Change an order** : See [Change a standard Provisioned Throughput order](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/purchase-provisioned-throughput#change-order). - **Schedule a new order** : Set a start date and time when you create an order. See [Place a standard Provisioned Throughput order](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/purchase-provisioned-throughput#place-an-order). - **Schedule a change to an order** : Set a start date and time when you change an order. See [Change a standard Provisioned Throughput order](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/purchase-provisioned-throughput#change-order). - **Split an order** : Divide an active order into two orders. See [Split an order](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/purchase-provisioned-throughput#split-order). ## June 25, 2026 Announcement **Administrative correction to Gemini Online Inference API on Gemini Enterprise Agent Platform SLA** We made an administrative correction to the Gemini Online Inference API on Gemini Enterprise Agent Platform [Service Level Agreement (SLA)](https://cloud.google.com/vertex-ai/generative-ai/sla?e=48754805), addressing a clerical issue. ## June 24, 2026 Feature **AI security findings in Agent Platform** Viewing AI security findings and posture management summaries in Gemini Enterprise Agent Platform is [generally available](https://cloud.google.com/products#product-launch-stages). With this release, the **Security** dashboard introduces the **Top security findings** widget. Also, specific features within the AI security widgets are available in [Preview](https://cloud.google.com/products#product-launch-stages), including the following: - Vulnerability findings and threat monitoring for agent runtimes (such as Cloud Run) - Historical content violation trends (**Violations over time**) For more information, see [View security findings](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/view-security-findings). Feature **Model Armor for Agent Gateway in General Availability** You can enable Model Armor on an Agent Gateway resource to apply your organization's content security guardrails to prompts and responses that pass through the gateway. This feature is generally available ([GA](https://cloud.google.com/products#product-launch-stages)). For more information, see [Configure Model Armor on a gateway](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/configure-model-armor). ## June 22, 2026 Change **Supervised fine-tuning available for Gemini 3.1 Flash Lite and Gemini 3.5 Flash in Public Preview** Supervised fine-tuning is now available for the `gemini-3.1-flash-lite` and `gemini-3.5-flash` models in [Preview](https://cloud.google.com/products#product-launch-stages). Model tuning for Gemini 3.1 Flash Lite and Gemini 3.5 Flash is restricted to `us-central1` and `europe-west4` and tuned model serving is restricted to the `us` and `eu` multi-region endpoints. See [About supervised fine-tuning](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini-supervised-tuning) for more information. Feature **Provisioned Throughput support for supervised fine-tuned Gemini 3 model inference.** Provisioned Throughput can be used to assure supervised fine-tuned inference using the same quota. Supervised fine-tuned inference for Gemini 3 models incurs a higher burndown rate compared to base model inference. Learn more [here](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/supported-models.md.txt#supervised-fine-tuned-model-support). ## June 18, 2026 Feature **Agent Gateway in General Availability** Agent Gateway is the networking component of the Gemini Enterprise Agent Platform ecosystem. It secures and governs connectivity for all agentic interactions, whether they occur between users and agents, agents and tools, or among agents themselves. For details, see [Agent Gateway overview](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/gateways/agent-gateway-overview). Feature **Agent Observability is generally available (GA)** This release provides visibility into the performance, behavior, and health of deployed agents and Model Context Protocol (MCP) servers directly within the agent management workflow. Key updates in this release include: - **Default-On Tracing:** OpenTelemetry tracing is now enabled by default for newly deployed Agent Development Kit (ADK) agents on Agent Engine, simplifying the observability setup process without requiring manual configuration. - **Storage Prioritization:** Google Cloud Storage (GCS) is the default storage choice in the Google Cloud Console, instead of Cloud Logging. We recommend that you store your multimodal prompt and response payloads in a Cloud Storage (GCS) bucket. This solution provides robust support for large payloads and it enables fine-grained lifecycle management. - **Enhanced Tracing:** Inspect step-by-step session execution and view directed acyclic graphs (DAGs) of trace spans. For more information, see the following: - [Observability overview](https://docs.cloud.google.com/gemini-enterprise-agent-platform/optimize/observability/overview.md.txt) - [View agent traces](https://docs.cloud.google.com/gemini-enterprise-agent-platform/optimize/observability/traces.md.txt) - [Set up tracing](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/runtime/tracing.md.txt) Feature **Agent Registry is generally available (GA)** Agent Registry is [generally available (GA)](https://cloud.google.com/products#product-launch-stages). Agent Registry is a centralized catalog for discovering and registering agents and Model Context Protocol (MCP) servers. The following features are available in Agent Registry for the GA launch stage: - **API v1 and client libraries:** The `v1` version of the Agent Registry API is available. Cloud client libraries are available in C#, Go, Java, Node.js, PHP, Python, and Ruby. - **A2A v1 support:** Agent Registry supports Agent-to-Agent (A2A) protocol version `1.0`, letting you explicitly declare transport endpoints and bindings inside the `supportedInterfaces` array, in addition to the existing `0.3` schema support. - **Terraform support:** Terraform scripts for Application Default Credentials (ADC) have graduated to General Availability. You can use Terraform to configure and manage your agents, MCP servers, endpoints, and bindings. **Known limitations:** - **Access Transparency and Access Approval:** [Access Transparency](https://docs.cloud.google.com/assured-workloads/access-transparency/docs/overview.md.txt) logs, which provide visibility into when Google personnel access your content, and [Access Approval](https://docs.cloud.google.com/assured-workloads/access-approval/docs/overview.md.txt) controls aren't available for Agent Registry configurations. - **Data Residency:** If you configure the [resource location constraint](https://docs.cloud.google.com/organization-policy/restrict-locations.md.txt) in your organization policy, Agent Registry enforces the constraint when you register a resource. However, detective controls for data residency compliance reporting are limited. For more information, see the [Agent Registry overview](https://docs.cloud.google.com/agent-registry/overview.md.txt). Feature The Agent Identity API (`agentidentity.googleapis.com`) is available in [Preview](https://cloud.google.com/products#product-launch-stages). This new API replaces the legacy IAM Connectors API (`iamconnectors.googleapis.com`) for managing auth providers and agent identities. During the preview migration period, both APIs operate side-by-side. Existing auth providers are automatically mirrored to the new V2 resource hierarchy (`authProviders/`), allowing you to migrate your IAM policies, agent code, and client applications without downtime. ## June 17, 2026 Feature **Memory Bank and Sessions global and multi-regional endpoints GA** Memory Bank and Sessions support for multi-regional and global endpoints is now in General Availability (GA). For more information, see [Supported locations for agents](https://docs.cloud.google.com/gemini-enterprise-agent-platform/resources/agent-locations.md.txt#multi-regional-and-global-endpoints). Note that Customer-Managed Encryption Keys (CMEK) cannot be used if your Memory Bank or Sessions instance is configured to use the global endpoint. ## June 15, 2026 Feature **Reinforcement Learning fine-tuning is available for Gemini 3.5 Flash in Public Preview** Reinforcement Learning fine-tuning is now available for the `gemini-3.5-flash` models in [Preview](https://cloud.google.com/products#product-launch-stages). Model tuning for Gemini 3.5 Flash is restricted to `us-central1` and `europe-west4`, and tuned model serving is restricted to the `us` and `eu` multi-region endpoints. See [About reinforcement learning fine-tuning](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/tuning/reinforcement-tuning.md.txt) for more information. ## June 09, 2026 Feature **Anthropic's Claude Fable 5** [Claude Fable 5](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude/fable-5.md.txt) is available in Model Garden. ## June 02, 2026 Announcement **Updates to abuse monitoring and zero data retention documentation** Documentation for abuse monitoring, zero data retention, and responsible AI has been updated to align with the Advanced AI Safety Addendum. These updates include new details regarding Advanced AI safety, partner-specific terms, and request-response logging for models like Claude Mythos and Opus. For more information, see: - [Abuse monitoring](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/abuse-monitoring.md.txt) - [Zero data retention](https://docs.cloud.google.com/gemini-enterprise-agent-platform/resources/zero-data-retention.md.txt) - [Responsible AI](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/responsible-ai.md.txt) ## June 01, 2026 Breaking **Gemini 2.0 Flash and Gemini 2.0 Flash-Lite are discontinued** Gemini 2.0 Flash and 2.0 Flash-Lite are discontinued and are no longer available. This includes both model serving and Provisioned Throughput. Use Gemini 3.1 Flash-Lite, Gemma 4, or more recent Gemini releases. ## May 28, 2026 Feature **Agent Platform Gemini 3.1 Flash Image and Gemini 3 Pro Image** Gemini Enterprise Agent Platform Gemini 3.1 Flash Image and Gemini 3 Pro Image are [Generally Available](https://cloud.google.com/products#product-launch-stages). With this release, Gemini 3.1 Flash Image and Gemini 3 Pro Image support 4K image outputs in [Preview](https://cloud.google.com/products#product-launch-stages). Also supported in this release, Gemini 3.1 Flash Image supports video inputs in [Preview](https://cloud.google.com/products#product-launch-stages). You can use video inputs to generate thumbnails or representative images of videos. For more information, see the following: - [Gemini 3.1 Flash Image (Nano Banana 2)](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-1-flash-image) - [Gemini 3 Pro Image (Nano Banana Pro)](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-pro-image) Deprecated **Agent Platform Gemini 3.1 Flash Image Preview and Gemini 3 Pro Image Preview deprecation** Gemini Enterprise Agent Platform Gemini 3.1 Flash Image Preview and Gemini 3 Pro Image Preview are deprecated. We recommend that you update your model endpoints before July 17, 2026, to avoid service disruption. The following are the discontinued endpoints and recommended endpoint migration: | Discontinued endpoints | Recommended endpoint migration | |---|---| | `gemini-3.1-flash-image-preview` | `gemini-3.1-flash-image` | | `gemini-3-pro-image-preview` | `gemini-3-pro-image` | Feature **Anthropic's Claude Opus 4.8** [Claude Opus 4.8](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude/opus-4-8.md.txt) is available in Model Garden. ## May 27, 2026 Feature **User ID logging now included with agent logs when you opt in to "Enable logging of prompt inputs and response outputs"** Prompt input and response output logging now includes the `user.id` field. This addition allows better tracking of anomalous tool interactions. > [!IMPORTANT] > **Important:** Logging of `user.id` is included when opting in to "Enable logging of prompt inputs and response outputs" effective May 22, 2026 and later and with the Agent Development Kit (ADK) version 2.1 and later. If you opted in prior to this change, your logs do not include `user.id`. You will need to redeploy your agents and opt-in again for this setting to take effect. For details on configuration, see [Write traces for an agent](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/runtime/tracing.md.txt#write-traces). ## May 26, 2026 Feature **The Gemini Deep Research Agent released in Preview** The Gemini Deep Research Agent has been released in Preview. The Gemini Deep Research Agent is a managed AI agent that plans, executes, and synthesizes complex, multi-step research workflows across the public web and private enterprise data to generate comprehensive, cited reports. For more information, see [Use the Gemini Deep Research Agent](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/use-deep-research). Feature **Agent Platform Sandboxes** Additional Agent Platform [sandbox](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/sandbox.md.txt) features are now available: - **[Computer use](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/sandbox/computer-use.md.txt) (Preview)**: Enables agents to automate browser-based tasks within an isolated web browser environment. You can control the browser using the API or connect directly using the Chrome DevTools Protocol (CDP). - **[Custom container sandboxes](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/sandbox/custom-containers.md.txt) (Preview)**: Bring your own container (BYOC) to run custom workloads with specialized dependencies hosted in Artifact Registry. - **[Sandbox templates](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/sandbox/manage-templates.md.txt) (Preview)**: Define sandbox specifications as reusable templates relying on pre-warmed pools to facilitate rapid, reliable startups. - **[Sandbox snapshots](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/sandbox/manage-snapshots.md.txt) (Preview)**: Save the exact state of your sandbox environment (including dependencies and file systems) and restore it to a new sandbox. Feature **Identify the agents with the most content security violations** The **Security** dashboard displays the top 10 agents with the most content violations detected by Model Armor. The list shows the agent ID of each agent and the number of violations detected for that agent. For more information, see [Monitor content security](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/monitor-content-security). ## May 20, 2026 Change **Supervised fine-tuning available for Gemini 3.1 Flash Lite (Preview)** Supervised fine-tuning is now available for limited support for the `gemini-3.1-flash-lite` model. During this period, model tuning for Gemini 3.1 Flash Lite is restricted to `us-central1` and `europe-west4` and tuned model serving is restricted to the `us` and `eu` multi-region endpoints. See [About supervised fine-tuning](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini-supervised-tuning) for more information. Change **Set media resolution at a Part-level for data when using supervised fine-tuning** Supervised fine-tuning now supports `Part`-level `mediaResolution` declarations for images, videos, and PDFs. `Part`-level media resolution declarations also support the `MEDIA_RESOLUTION_ULTRA_HIGH` level. See the following media type--specific pages for more information: - [Document tuning](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/tune_gemini/doc_tune) - [Image tuning](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/tune_gemini/image_tune) - [Video tuning](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/tune_gemini/video_tune) ## May 19, 2026 Feature **Gemini 3.5 Flash is generally available (GA)** For details, see the [model specifications page](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-5-flash.md.txt). Feature **Manage agent revisions and traffic splitting** Agent revisions and traffic splitting are now available in public preview. You can create immutable revisions of deployed agents, and split traffic between the different active revisions. This enables canary deployments and safe testing of new agent versions. For more information, see [Manage revisions and traffic](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/runtime/manage-revisions-and-traffic.md.txt). Feature **Manage and discover agent skills with Skill Registry** Manage and discover agent skills with the Skill Registry, in public preview. This secure, private, and low-latency repository stores skills as self-contained packages, including instructions, code, and documentation, to enhance agent abilities. For more information, see: - [Skill Registry overview](https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/skill-registry.md.txt) - [Create and manage skills](https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/skill-registry/create-manage.md.txt) Feature **Managed Agents API on Agent Platform released in Preview** The Managed Agents API on Agent Platform has been released in Preview. This feature allows you to build and scale autonomous agents, including those built from configuration using the Antigravity harness. These agents run in a fully managed and isolated sandbox environment, equipped with tools and skills, and can be interacted with via a dedicated API. For more information, see the following: - [Managed Agents API on Agent Platform overview](https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/managed-agents.md.txt) - [Create and manage agents](https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/managed-agents/create-manage.md.txt) - [Interact with agents](https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/managed-agents/interact-with-agents.md.txt) - [Managed Agents API on Agent Platform sandbox environment](https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/managed-agents/sandbox-environment.md.txt) Feature **AI Content Detection API available** AI Content Detection API is available in [Preview](https://cloud.google.com/products#product-launch-stages). For details see [AI Content Detection](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/ai-content-detection.md.txt). Feature **Provisioned Throughput for Gemini now supports latency SLA** Provisioned Throughput now provides a tokens per second latency SLA, covering generation speed from the first returned token to the last. For more information, see the [Gemini Enterprise Agent Platform Online Inference Service Level Agreement (SLA)](https://cloud.google.com/vertex-ai/generative-ai/sla?e=48754805). Change Memory Bank and Sessions support for multi-regional and global endpoints is in [Preview](https://cloud.google.com/products#product-launch-stages). For more, see [Supported locations for agents in Agent Platform](https://docs.cloud.google.com/gemini-enterprise-agent-platform/resources/agent-locations.md.txt#multi-regional_and_global_endpoints). ## May 14, 2026 Feature **Priority PayGo is generally available (GA)** Priority PayGo is a consumption option that provides more consistent performance than standard PayGo without the upfront commitment of Provisioned Throughput. It is ideal for business-critical workloads with fluctuating or unpredictable traffic patterns. For more information, see [Priority PayGo](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/priority-paygo.md.txt). ## May 11, 2026 Change You can purchase **Provisioned Throughput for Gemma 4** . To learn more, see the list of [supported open models](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/supported-models#open-models). ## May 08, 2026 Change **Gemini Distillation Service Early Access** We're introduction Gemini Distillation Service in Early Access. For information about requesting access, see [Gemini Distillation Service](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/tuning/distillation). Change **Improvements to the Provisioned Throughput orders page** have now made it possible to: - View all scheduled orders by using the Start Date column. - Enable filtering and sorting of orders by using column names. - Download all order data to a CSV file (including across all regions). See [View standard Provisioned Throughput orders](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/purchase-provisioned-throughput#view-orders). ## May 07, 2026 Change **Gemini 3.1 Flash-Lite is now generally available** Our most cost-efficient Gemini model, **3.1 Flash-Lite** , is out of preview and is now generally available. For technical information on this model, see the [model information card](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/3-1-flash-lite.md.txt). Change You can purchase **Provisioned Throughput for Gemini 3.1 Flash-Lite** . To learn more, see the [Provisioned Throughput overview](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/overview). ## May 06, 2026 Fixed Fixed an issue with **Audio track extraction (Gemini Embedding 2 only)** where the `audio_track_extraction` feature did not work. For more information, see [Issue #504505771](https://issuetracker.google.com/504505771). Change **Agent Platform Gemini 3.1 Flash Image and Gemini 3 Pro Image** Gemini Enterprise Agent Platform Gemini 3.1 Flash Image Preview and Gemini 3 Pro Image Preview are introducing the following changes: - Upgrades to improve 4K outputs and efficiency in both models - Gemini 3.1 Flash Image Preview and Gemini 3 Pro Image Preview will now return a maximum of 1 thought image. - The image_size parameter for Gemini 3.1 Flash Image Preview now accepts "512", "512p", "512P", "512PX", "512px" to generate 0.5MP resolution output images. - The default thinking level for Gemini 3.1 Flash Image Preview changed to Minimal. ## April 28, 2026 Feature **Improved transcription quality for Gemini Live API** You can now improve transcription quality for multilingual automatic speech recognition (ASR) by using the `[input/output]_audio_transcription.language_codes` field. For more information, see [Enable audio transcription for the session](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/live-api/start-manage-session.md.txt#enable-audio-transcription). Feature **Asynchronous function calling with Live API** Asynchronous function calling is now available in [public preview](https://cloud.google.com/products#product-launch-stages) in Gemini Live API. You can run functions in parallel with conversation, manage background processing, and handle function responses with policies including `SILENT`, `WHEN_IDLE`, and `INTERRUPT`. For more information, see [Asynchronous function calling with Gemini Live API](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/live-api/asynchronous-function-calling). ## April 22, 2026 Change **Vertex AI to Gemini Enterprise Agent Platform naming changes** The table below lists all of the features that have been transitioned from Vertex AI and what their new names are in Agent Platform. #### Click to expand naming changes list | Vertex AI name | Agent Platform name | |---|---| | Vertex AI Platform | Agent Platform | | Generative AI on Vertex AI | Generative AI | | Vertex AI Studio | Agent Studio | | Vertex AI API | Agent Platform API | | Vertex AI Model Garden | Model Garden | | Vertex AI Models as a Service (MaaS) | MaaS | | (Gemini/Veo) on Vertex AI | (Gemini/Veo) on Agent Platform | | (Claude/Llama/DeepSeek/etc.) on Vertex AI | (Claude/Llama/DeepSeek/etc.), available on Agent Platform | | Pre-trained APIs on Vertex AI | Pre-trained APIs on Agent Platform | | (Provisioned Throughput/Pay-as-you-go/etc.) on Vertex AI | (Provisioned Throughput/Pay-as-you-go/etc.) on Agent Platform | | Gemini Live API on Vertex AI | Gemini Live API on Agent Platform | | Vertex AI Search | Agent Search | | Vertex AI Search for Industry | Agent Search for Industry | | Vertex AI Search for Commerce | Agent Search for Commerce | | Recommendations from Vertex AI Search | Recommendations | | Vertex AI Conversation | Agent Conversation | | Vertex AI RAG Engine | RAG Engine | | Vertex AI Vector Search | Vector Search | | Vertex AI Vector Search 2.0 | Agent Retrieval | | Vertex AI Agent Engine | Agent Runtime | | Vertex AI Studio App Builder | App Builder in Agent Studio | | Vertex AI Agent Engine Memory Bank | Agent Platform Memory Bank | | Vertex AI Agent Engine Sessions | Agent Platform Sessions | | Vertex AI Agent Engine Code Execution | Agent Platform Code Execution | | Grounding with Google \[...\] in Vertex AI | Grounding with Google \[...\] in Agent Platform | | Grounding with Google \[...\] in Vertex AI Search | Grounding with Google \[...\] in Agent Search | | Grounding with Google \[...\] in Vertex AI Studio | Grounding with Google \[...\] in Agent Studio | | Vertex AI Training | Agent Platform Managed Training | | Vertex AI Serverless Training | Agent Platform Serverless Training | | Vertex AI Training Clusters (VTC) | Managed Training Clusters | | Ray on Vertex AI | Ray on Agent Platform | | Reinforcement Learning from Human Feedback (RLHF)/Reinforcement Learning (RL) on Vertex AI | Reinforcement Learning on Agent Platform | | Vertex AI Neural Architecture Search | Neural Architecture Search on Agent Platform | | Vertex AI Prediction/Vertex AI Inference | Agent Platform Inference | | Vertex AI Vision | Agent Platform Vision | | Vertex AI Batch Inference | Agent Platform Batch Inference | | Vertex AI Online Inference | Agent Platform Online Inference | | Vertex AI Endpoints | Agent Platform Endpoints | | Vertex AI Forecasting/Forecasting with AutoML | Forecasting on Agent Platform | | Vertex AI Pipelines | Agent Platform Pipelines | | Vertex AI Notebooks | Agent Platform Notebooks | | Vertex AI Colab Enterprise | Agent Platform Colab Enterprise | | Vertex AI Workbench | Agent Platform Workbench | | Vertex AI Workbench Instances | Agent Platform Workbench Instances | | Vertex AI Feature Store | Agent Platform Feature Store | | Vertex AI Model Registry | Agent Platform Model Registry | | Vertex AI Model Evaluation | Agent Platform Model Evaluation | | Gen AI evaluation service on Vertex AI | Gen AI evals | | Vertex AI AutoML (Vision/Video/Tables) | Agent Platform AutoML | | Data Labeling on Vertex AI | Data Labeling | | Vertex AI on GDC | Agent Platform on GDC | | Vertex AI Experiments | Experiments on Agent Platform | | Vertex AI Model Monitoring | Model Monitoring on Agent Platform | | Vertex AI Media Studio | Agent Media Studio | | Vertex AI | Agent Platform | | Vertex AI Generative AI | Agent Platform Generative AI | Change **Initial release of Gemini Enterprise Agent Platform** This initial release includes (but is not limited to) the following releases or changes: - Change **Vertex AI** is now part of Gemini Enterprise Agent Platform. Information on model support for Vertex AI is now under [Gemini Enterprise Agent Platform \> Models](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/overview). - Change **Agent Builder** is now part of Gemini Enterprise Agent Platform. Features have been renamed as follows: - **Agent Engine** is now [**Agent Runtime**](https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/runtime.md.txt). - **Agent Builder Sessions** is [**Agent Platform Sessions**](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/sessions.md.txt). - **Memory Bank** is now [**Agent Platform Memory Bank**](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/memory-bank.md.txt). - Change **Agent Runtime** now supports long-running operations (up to 7 days). - Change **Agent Runtime** now supports sub-second cold starts. - Change Provisioning for **Agent Runtime** has been reduced to less than 1 minute. - Release You can now use your own [custom-built containers](https://docs.cloud.google.com/gemini-enterprise-agent-platform/release-notes/gemini-enterprise-agent-platform/build/runtime/setup#byoc) when you deploy agents with **Agent Runtime**. - Change When [creating a Session](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/sessions/manage-with-api), you can specify your own session ID. - Release Memory Bank now enables continuous event streaming with automated memory generation triggered by configurable criteria like event count or idle time. For more information, see [Ingest events](https://docs.cloud.google.com/gemini-enterprise-agent-platform/scale/memory-bank/ingest-events). - Release Memory Bank now automatically maintains an immutable version history of memories through revision resources. For more information, see [Memory revisions](https://docs.cloud.google.com/gemini-enterprise-agent-platform/release-notes/gemini-enterprise-agent-platform/scale/memory-bank/revisions). - Release [**Agent Identity**](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/agent-identity-overview) for General Availability. Agent Identity helps let your agent securely authenticate to MCP servers, cloud resources, endpoints, and other agents, either acting as itself or acting on behalf of the end user. - Release [**Agent Gateway**](https://docs.cloud.google.com/gemini-enterprise-agent-platform/govern/gateways/agent-gateway-overview.md.txt) for Private Preview. Agent Gateway is the networking component of the Gemini Enterprise Agent Platform ecosystem. It secures and governs connectivity for all agentic interactions, whether they occur between users and agents, agents and tools, or among agents themselves. - Release [**Agent Registry**](https://docs.cloud.google.com/agent-registry/overview) for Public Preview. Agent Registry is a centralized, unified catalog that lets you store, discover, and govern Model Context Protocol (MCP) servers, tools, and AI agents within Google Cloud. - Release New [**IAM governance policies**](https://docs.cloud.google.com/gemini-enterprise-agent-platform/release-notes/gemini-enterprise-agent-platform/govern/policies/overview) are available in Private Preview. - Release [**Agent Observability**](https://docs.cloud.google.com/gemini-enterprise-agent-platform/optimize/observability/overview) for Preview. Agent Observability in Gemini Enterprise Agent Platform provides comprehensive visibility into the performance, behavior, and health of your deployed agents and Model Context Protocol (MCP) servers. By monitoring key metrics, tracing execution paths, and observing your multi-agent system as a whole, you can diagnose issues, optimize resource consumption, and improve the reliability of your agents. - Release [**Gemini Embedding 2**](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/gemini/embedding-2) (`gemini-embedding-2`) for General Availability. - Release The [**Gemini Deep Research Agent**](https://docs.cloud.google.com/gemini-enterprise-agent-platform/agents/deep-research), a pre-built agent designed to help you plan, execute, and synthesize multi-step research tasks. It uses Gemini 3.1 Pro to bridge the gap between public web data and private enterprise context by simultaneously grounding research across three distinct, high-fidelity data streams. - Release [**Agent Platform remote MCP server**](https://docs.cloud.google.com/gemini-enterprise-agent-platform/reference/use-agent-platform-mcp) for General Availability. Support for Model Context Protocol (MCP) use is available for Agent Platform. - Change **Google Cloud console navigation**: The navigation menus under Agent Platform (formerly Vertex AI) and Data Analytics have been updated to centralize agentic products and features. Bookmarked links will continue to work via automatic redirects. Issue The following known issues affect Gemini Enterprise Agent Platform: - **Audio track extraction (Gemini Embedding 2 only):** The `audio_track_extraction` feature does not work. For more information, see [Issue #504505771](https://issuetracker.google.com/504505771). ## March 31, 2026 Security **Security update for Cloud Storage staging bucket verification** A security issue in the Python SDK (`google-cloud-aiplatform`) for Gemini Enterprise Agent Platform allowed potential Cloud Storage bucket squatting when staging local model artifacts without specifying an explicit staging bucket. This issue has been resolved by adding bucket ownership verification and randomized naming for default staging buckets in Cloud Storage. Users of the `google-cloud-aiplatform` SDK should upgrade to version 1.165.1 or later, or ensure that an explicit `staging_bucket` belonging to their project is specified when calling `Model.upload()`.