← Model indexMODEL PROFILE / claude-opus-5-5

Anthropic / current

Claude Opus 5.5

Anthropic's current flagship for agentic coding and knowledge work.

Context1M
ModalityText · Image
AccessAPI · Cloud
ReasoningAdaptive

Where it fits.

Opens the Claude 5.5 family, matching most Fable 5.1 quality while cutting typical run cost by roughly forty percent. Released September 22, 2026 as the first Claude 5.5 model, positioned for agentic coding and knowledge work at roughly 40% lower cost than Opus 5 on typical workloads.

02

Core capabilities

  • 1M context window with up to 128K output
  • Always-on adaptive thinking for long-horizon work
  • Agentic tool use with preserved thinking blocks
  • Availability through Claude API and major managed clouds
03

Best-fit workflows

  • Long-running agents
  • Complex refactoring
  • Enterprise knowledge work
04 / OPERATIONS

From evaluation
to production.

Deployment notes

Adopt as the escalation target above Sonnet 5.5. Return thinking blocks unmodified across tool calls, checkpoint long runs, and compare against Sonnet 5.5 on a task-specific eval.

Watch before choosing

Premium tier pricing still applies; validate the quality uplift against your own evals before switching wholesale. Cache-read pricing and effort settings materially change effective cost; benchmark at your real context length.

05

Family & versions

Specifications and positioning checked against provider documentation.

Open official source↗