What happened: Anthropic released Claude Opus 4.7 on April 16, 2026 — its most capable generally available model to date and a notable course-correction after the controversial Claude Mythos preview the same week.
Opus 4.7 lifts performance across the board: SWE-bench Verified hit 87.6%, software engineering tasks are more reliable with fewer tool errors, and instruction-following improved 10–15% on factory task benchmarks. It replaces Opus 4.6, which launched in February.
Why it matters: Anthropic clearly heard the feedback on Mythos. Opus 4.7 is being positioned as the “production-safe” alternative — still cutting-edge but with the guardrails enterprises need. The Mythos safety questions triggered a course correction that is now reflected in a model designed for real deployment, not demo sessions.
For developers: Opus 4.7 is available via the Anthropic API now. Cloudflare Docs already lists it as a supported model. If you were on Opus 4.6, the upgrade path is straightforward — no prompt engineering changes required.
The gap between Anthropic and OpenAI/Gemini continues to narrow in specific benchmarks, but Opus 4.7 keeps them firmly in the race on coding and complex reasoning tasks.
If you spot a factual error, drop a comment — I respond and correct.
