Claude Opus 5.5

Claude Opus 5.5 is built for long-running agentic coding and knowledge work. Anthropic positions it at roughly Fable 5.1 intelligence for most tasks, at $4/$20 per million tokens instead of Fable's $10/$50, and it leads Anthropic's lineup on agentic coding benchmarks such as Terminal-Bench 4.0 (66.4%) and CursorBench 4.0 (57.8%). Adaptive thinking is always on, with depth controlled by an effort setting that defaults to medium. It accepts text and images with a 1M-token context window and returns up to 128K output tokens (300K via the Batch API).