We've launched Claude Haiku 5.5 (claude-haiku-5-5), our most capable model tuned for high-volume and latency-sensitive work.
What to do: Review the migration guide before switching existing Claude Haiku 4.5 integrations to Claude Haiku 5.5.
News wire archive
Every item on the wire, newest first. Times are Eastern (ET).
WednesdayOctober 7, 2026
We've lowered the price of prompt cache reads on Claude Sonnet 5.5 from $0.20 USD to $0.10 USD per million tokens: 0.05x the base input price instead of 0.1x.
What to do: Review your prompt caching costs on Claude Sonnet 5.5 to capture the reduced read price; no code changes are needed.
TuesdayOctober 6, 2026
Gemini Nano Banana 2.1 generally available (GA) : Released Gemini Nano Banana 2.1 (gemini-nano-banana-2.1), the latest high-efficiency image generation and con…
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.
Released the Decisions API in beta with gpt-6-luna.
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.
Simplified API usage tiers from five to three: Build, Launch, and Grow.
What to do: Review the usage tiers documentation to see the new monthly limits and rate limits for your organization's tier.
python - v3.1.0 - 2026-10-06 15:53:07
What to do: Upgrade to the Mistral Python SDK v3.1.0 on PyPI and update code that uses the removed Runs/Records methods or changed authentication and pipeline types.
Introducing Mistral Large 4
What to do: Try the Mistral Large 4 preview API in Mistral Studio and plan for open-weight availability by end of the month.
We've added capabilities.server_tools to the Models API.
What to do: Read capabilities.server_tools from the Models API endpoints to check tool support per model instead of hardcoding it.
MondayOctober 5, 2026
Added an in-product flow for HIPAA compliance support in API Organization settings > General.
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.
We've added capabilities.thinking.types.disabled to the Models API.
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"}.
SaturdayOctober 3, 2026
InfoHacker News · community
Getting the most out of Opus 5.5 in Claude and Claude Code
What to do: Review your prompts and CLAUDE.md files to remove 'think step by step' style instructions and add stop/continue rules before running long tasks on Opus 5.5.
InfoHacker News · community
Show HN: Offrun – manage every coding agent from one workspace
FridayOctober 2, 2026
ThursdayOctober 1, 2026
InfoGoogle
How to Evaluate Live & Voice Agents in ADK
What to do: Install the ADK eval extras with uv pip install -e ".[eval]" and configure API credentials for both the Live API and Gemini TTS to enable live voice agent evaluation.
InfoGoogle
Build zero-trust AI agents that judge intent, not just syntax
What to do: Review the new zero-trust runtime governance controls (Model Armor, Semantic Governance Policies, and Agent Anomaly Detection) if you deploy agents to the Gemini Enterprise Agent Platform, and consid…
Agent Anomaly Detection, now in Private Preview on the Gemini Enterprise Agent Platform
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.
Introducing Support for Local AI Models in the Antigravity SDK
What to do: Try local agentic workflows with the Antigravity SDK by following the README instructions, using a machine with >24GB VRAM or unified memory for Gemma 4 26B A4B via LiteRT, or connect your own OpenAI…
InfoGoogle
Turn your REST APIs into MCP tools with Google Cloud API Gateway
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.
We've added a line field to the Models API.
What to do: Use the new line field to group models instead of parsing model IDs.