Every change we detect, newest first. Times are Eastern (ET). Archive
What to do: If you use limited networking with web tools, add any hosts in allowed_domains to allowed_hosts (or remove those entries) to avoid 400 errors on session creation.
What to do: If you use the Compliance API for Claude Enterprise, review the updated chat retrieval docs to include unified Claude experience chats.
What to do: Review the migration guide and updated pricing, and consider switching high-volume workloads to claude-haiku-5-5 and updating SDKs for the new computer/browser use features.
What to do: Use the GPT-6 model family for production workloads and reserve chat-latest for testing, since its snapshot changes regularly.
What to do: Review the migration guide before switching existing Claude Haiku 4.5 integrations to Claude Haiku 5.5.
What to do: Migrate image generation code from gemini-3.1-flash-image to gemini-nano-banana-2.1 before the October 29, 2026 shutdown.
What to do: Try the beta Decisions API at v1/decisions with gpt-6-luna if you need faster typed answers from text or images.
What to do: Review the usage tiers documentation to see the new monthly limits and rate limits for your organization's tier.
What to do: Try the Mistral Large 4 preview API in Mistral Studio and plan for open-weight availability by end of the month.
What to do: Read capabilities.server_tools from the Models API endpoints to check tool support per model instead of hardcoding it.
What to do: If your organization needs HIPAA compliance, have an admin enable it in API Organization settings > General and review the Help Center eligibility and configuration requirements.
What to do: Update model capability checks to read capabilities.thinking.types.disabled from the Models API when deciding whether to send thinking: {type: "disabled"}.
What to do: No action needed; background reading.
What to do: If you deploy agents on the Gemini Enterprise Agent Platform with ADK 1.2 or later, review the documentation for prerequisites and setup to enable Agent Anomaly Detection with one-click provisioning.
What to do: Annotate your OpenAPI spec with MCP metadata and deploy it to Google Cloud API Gateway to expose existing REST APIs as MCP tools without building a separate server.
What to do: Use the new line field to group models instead of parsing model IDs.
What to do: Watch for the wider release to paid API customers and plan for the announced pricing of $2 per million input tokens, $10 per million output tokens, and cached input at 95% off.
What to do: Try the open-source Weave Router if you want to reduce costs and latency for coding agents by routing between multiple LLMs instead of using a single frontier model.
What to do: Check whether this affects your integration and plan for it.
What to do: Use service_tier: "ultrafast" with gpt-6-astra in the Responses API to reduce inter-token latency, keeping in mind EU and other regional inference residency are not supported.
What to do: Use the Responses API for tool calling and see GPT-6 model guidance for reasoning settings and pricing for available processing tiers.
What to do: No action needed; background reading.
What to do: Review the migration guide and update code that targets Claude Sonnet 5 before using Claude Sonnet 5.5, especially for tool_choice, thinking configuration, computer use tool names, and advisor model…
What to do: Rerun evaluations and retry affected workflows if your use cases involve image inputs.