Claude Opus 4.1

Claude Opus 4.1 (API name claude-opus-4-1-20250805) shipped on 5 August 2025 as an incremental upgrade to Opus 4, aimed squarely at agentic coding and multi-step research. It was a hybrid reasoning model: near-instant answers by default, or extended thinking with a developer-set budget when a task warranted it, inside a 200,000-token context window with up to 32,000 output tokens. Anthropic priced it at $15 per million input tokens and $75 per million output, the standard Opus rate of that generation, and released it under ASL-3 deployment safeguards. Its headline number was 74.5% on SWE-bench Verified, then a leading score for real-world software-engineering tasks. The reason this profile exists is the date at the other end. Anthropic notified developers on 5 June 2026 that the model would be retired, and it was switched off on 5 August 2026 — exactly one year after it launched. Requests to it now fail; the recommended replacement is claude-opus-4-8. That one-year span is the honest headline for how fast this segment moves: a model that was, briefly, the best coding model available anywhere is now unreachable, and the entire arc from flagship to dead endpoint fits inside twelve months. Anthropic has committed to preserving the weights of retired models rather than deleting them, and has said it hopes to make past models publicly available again — a commitment that matters to researchers doing comparative work, who otherwise lose their baselines every time a vendor cleans house. wujec.ai keeps profiles of superseded and retired models. This one is history now, not a buying decision.

▸Videos

No videos yet.

▸Reviews