LLM Tech documents a capability change: “2026-09-18 Service is back after the maintenance window, on new hardware: one RTX PRO 6000 Blackwell with 96 GB, in Italy.”
Stay ahead of the change.
Releases, retirements, and the deadlines that affect your stack.
Track the models you actually use.
Build a local watchlist to see changes that affect your application.
Next 3 model shutdowns
Select a model to review its notice and affected scope.Change activity
Published announcements · Jan 2026 – Dec 2026
Mixed days use a separate shade; hover for the counts or select a day to explore its notices. Undated announcements stay in the feed; empty days do not imply complete coverage.
Latest model changes
4Tracking since Jan 1, 2026 · recent announcements and deadlines from that date onward.
1189 published notices · 2,611 observations preserved in the archive. Older, undated, or unsupported claims can be hidden.
LLM Tech documents an alias change: “faster than the same prompt cold.”
LLM Tech announced this model on 2026-08-22.
LLM Tech announced this model on 2026-08-22. serving live marketplace traffic on NanoGPT.
Early signals & rumors 18
Reports of what might be next, with the original source attached.
Is Space Bunny Alpha MiniMax? Evidence Review
Kiana Qwen Mystery Model on Code Arena: What We Know | Payload Website Template
Curated AI Timeline for: 'Reasoning Model' | Annielytics
Open Source & Open Weights: AI Releases & Expert Coverage — ThursdAI
Split-boundary leaks in streaming LLM guardrails
Leaked Gemini 4 Pro Benchmarks Signal Google's Return to the Top of AI
Unconfirmed reports stay separate from official notices, countdowns and feeds.
Dates come from official notices. Missing dates stay missing.