Anthropic Python SDK 1.15.0
What to do: Update to SDK v1.15.0 if you need the new speed parameter or spend limit fields, and check whether your model requires a beta header for speed.
AIWire: AI model shutdowns, prices and API changes
The wire
Every change we detect, newest first. Times are Eastern (ET). Archive
ThursdayOctober 15, 2026
The Compliance API now returns the text each organization member saved as Instructions for Claude in claude.ai and as Global instructions in Cowork, in beta fo…
What to do: If you use the Compliance API for Claude Enterprise, call the new user_preferences endpoint with your Compliance Access Key and the read:compliance_user_data scope.
NVIDIA deprecated nemotron-3-super-120b-a12b: shuts down Oct 19, 2026
What to do: Plan a move before Oct 19, 2026.
Announcing DPoP Support for Google OAuth: Securing Refresh Tokens
What to do: Review the DPoP adoption guide and SKILL.md to add DPoP binding for refresh tokens in your Google OAuth integrations.
Muse Spark Open
InfoReleasesHacker News · community
What to do: Review the Muse Model License terms and try the model via a hosted inference partner or download the weights from Hugging Face.
NVIDIA added nemotron-3.5-super-vl-120b-a12b
What to do: Call it as nemotron-3.5-super-vl-120b-a12b.
Evaluating AI agents: moving past the vibe check
What to do: Consider adopting a staged eval pipeline (synthetic scenarios, red-pen trace review, then CI/CD gates) and the agent-eval toolkit to catch silent agent regressions before shipping.
Mistral added timestral-1-0-5m
What to do: Not on the API; download the weights to run it yourself.
Mistral added timestral-1-0-28m
What to do: Not on the API; download the weights to run it yourself.
Mistral added timestral-1-0-123m
What to do: Not on the API; download the weights to run it yourself.
Mistral added timestral-1-0
What to do: Not on the API; download the weights to run it yourself.
Introducing Timestral
What to do: Try Timestral via the GitHub repository if you need CPU-based time-series forecasting, or use Mistral Forge for domain fine-tuning.
We've added capabilities.thinking.types.between_tools to the Models API.
What to do: Remove the fast-mode-2026-02-01 beta header from Claude Opus 5.5 fast mode requests if you no longer need it, and check the Models API for between_tools thinking support.
WednesdayOctober 14, 2026
Google Gen AI Python SDK 2.31.0
What to do: Update the google-genai Python SDK to v2.31.0 to pick up the new model enum entry and continuation/grounding fixes.
Announcing MCP support in Apigee hybrid: agentic tools that run in your own cluster
What to do: If you run Apigee hybrid 1.17.0 or later and want agentic API access, enable MCP via Helm in a maintenance window, upgrading the apigee-operator chart before the apigee-org chart.
We've launched OpenTelemetry export in beta, so you can trace Messages API requests and Claude Managed Agents sessions in your observability backend.
What to do: Add an OpenTelemetry destination in the Claude Console and instrument your SDK calls to start tracing requests.
Anthropic Python SDK 1.14.0
What to do: Update to v1.14.0 to get OpenTelemetry tracing and the streaming fixes.
Grok Voice 2.0 deprecation
What to do: Update your voice model setting from grok-voice-2.0 to grok-voice-3.0 or grok-voice-latest.
Grok Voice 3.0 is available
What to do: No action needed unless you want to pin a specific version; otherwise sessions on grok-voice-latest upgrade automatically.
Google raises gemini-4-argon prices from Dec 1, 2026: input $2.00 → $4.00, output $10.00 → $20.00 per 1M tokens; batch, flex and priority tiers too
What to do: Check your cost estimates before Dec 1, 2026.
gemini-2.5-flash-preview-tts: Batch API supported → not supported
What to do: Check requests that use Batch API with gemini-2.5-flash-preview-tts.
OpenAI Python SDK 3.31.0
What to do: Update the OpenAI Python SDK to v3.31.0 to use the new browser and computer toolsets.
OpenAI Python SDK 3.30.0
What to do: Update the OpenAI Python SDK to v3.30.0 to pick up the new API features and the streaming connection-reuse fix.
We've launched OpenTelemetry export in beta, so you can trace Messages API requests and Claude Managed Agents sessions in your observability platform.
What to do: Add an OpenTelemetry destination in the Claude Console to start receiving spans, and review your organization's web search domain restrictions.
TuesdayOctober 13, 2026
Show HN: Parsimony – an open-source benchmark for how much code AI models add
InfoResearchHacker News · community
What to do: Check the Parsimony leaderboard when picking a model for coding agents if code bloat is a concern.
Gemini 4 Argon (gemini-4-argon) generally available (GA) : Released Gemini 4 Argon (gemini-4-argon), our most capable model built for complex coding, agentic,…
What to do: Review the What's new in Gemini 4 Argon guide for API changes, pricing, and migration guidance before adopting the new model.
Optimizing a 200,000-line Python codebase overnight with Gemini 4 Argon
What to do: Review the model card and developer guide for gemini-4-argon specifications, token limits, and migration instructions, and consider adopting the .agents/skills plus /goal overnight workflow on your o…
Gemini 4 Argon for Robotics
What to do: Explore Gemini 4 Argon on Google's developer platforms if you're building robotics or physical-reasoning applications.
Gemini 4 Argon: Built to solve your hardest problems
What to do: Review the Gemini API documentation for the new continuation protocol and try Argon in AI Studio if you need long-context, multi-step agentic workloads.