Track updates, pricing shifts, and deprecations across OpenAI, Anthropic, Google, Meta, Mistral, and more.
Free forever. No spam. Unsubscribe anytime.
847
Changes tracked
23
Repos monitored
6
Providers
4.2K
Subscribers
Live Feed
Latest changes
Real-time updates from every major AI provider. Severity-scored and summarized.
Breaking OpenAI2026-04-01
Deprecated gpt-4-32k and gpt-4-32k-0613
Both models will stop serving requests on June 1, 2026. Existing fine-tunes must be migrated to gpt-4o or gpt-4.1. No direct replacement with 32k context at the same price point.
New flagship reasoning model. 1M token context window, improved code generation, agentic tool use, and extended thinking. Available via API and Claude Pro/Max.
Gemini 2.5 Pro pricing increased 15% across all tiers
Input token pricing moves from $1.25 to $1.44 per million tokens. Output remains unchanged. Effective April 15, 2026. Gemini 2.5 Flash pricing unchanged.
Responses API replaces Chat Completions as default in SDK v5
openai-python v5.0 and openai-node v5.0 now default to the Responses API. Chat Completions still available but import paths changed. Migration guide published.
New mid-tier model slots between Small and Large. 32k context. Optimized for European language tasks. Available on Azure and AWS Bedrock within 2 weeks.
Code interpreter runs in a sandboxed environment by default. Supports Python, JavaScript, and shell. Tool use latency reduced 40% compared to GPT-4o with code interpreter.
All Gemini 1.0 Pro endpoints now return 404. Migrate to Gemini 2.5 Pro or Gemini 2.5 Flash. Projects using gemini-pro alias will be automatically routed to gemini-2.5-flash.
Claude Sonnet 4 added to AWS Bedrock cross-region inference
Sonnet 4 now available in US East, US West, and EU West regions via Bedrock. On-demand and provisioned throughput supported. Pricing matches direct API.