Claude Opus 5.5
Claude Opus 5.5 is Anthropic's September 22, 2026 model with the API ID claude-opus-5-5, a 1M-token context window, 128K maximum output, text and image input, tool use, and adaptive thinking that is always on.
Anthropic now recommends Opus 5.5 as the starting point for most workloads. It lowers input and output rates by 20% versus Opus 5 and cache-read rates by 60%, which can materially change the cost of long Claude Code sessions without making it the right choice for every unsupervised or frontier task.
Anthropic's release notes, model overview, pricing documentation, and cost article control identity, limits, positioning, and prices. The performance and task-cost statements are Anthropic's own measurements or illustrations; readers should verify them on their work.
- Supervised multi-file feature work.
- Long-running agentic coding sessions.
- Tool-using debugging and code review.
- Knowledge work with document and image input.
- Migration from Claude Opus 5.
The API model ID is claude-opus-5-5. Anthropic documents 1M context, 128K output, text and image input, text output, tool use, adaptive thinking always on, medium default effort, and a June 2026 reliable knowledge cutoff.
API list prices are $4 per million input tokens, $20 output, and $0.20 cache reads. Five-minute and one-hour cache writes are $5 and $8. Cache-heavy coding sessions can see a larger input-cost reduction than short uncached requests, but actual task cost also depends on turns and output.
Anthropic recommends starting most work on Opus 5.5 and moving to Fable 5.1 for demanding reasoning or long-horizon agent tasks when higher-effort Opus evals fall short. That is provider guidance, not a substitute for workload-specific acceptance tests.
Claude Opus 5.5 FAQ
Common questions about Claude Opus 5.5.
What is the Claude Opus 5.5 API model ID?+
Use claude-opus-5-5. Anthropic lists a 1M context window, 128K output, adaptive thinking, vision, and tool use.
How much does Claude Opus 5.5 cost?+
API list prices are $4 per million input tokens and $20 output, with cache reads at $0.20. Cache writes use separate five-minute and one-hour rates.
Should I use Opus 5.5 or Fable 5.1?+
Start with Opus 5.5 for most supervised work. Test Fable 5.1 when demanding reasoning or long-horizon agents still miss your acceptance bar at higher Opus effort.