Claude Opus 5.5
Claude Opus 5.5 is Anthropic's September 22, 2026 Opus-class model for long-running agentic coding and knowledge work, and the first model in the Claude 5.5 family.
Provider
Anthropic
Model family
Anthropic Claude
Multimodal LLM
Cost tier
Opus
Status
Current
Release Sep 22, 2026
Why teams choose it
List price is $4 input / $20 output per million tokens
Cache reads are $0.20, 5-minute cache writes $5, and 1-hour cache writes $8 per million tokens.
Anthropic says input and output tokens are 20% less than Opus 5, cache reads are 60% less, and output is more than 30% faster
Their tests put typical default-setting cost about 40% below Opus 5.
Fast mode is a separate research-preview rate at $8 input / $40 output per million tokens
up to 2.5× speed in Claude Code and the Claude Platform.
Tradeoffs to know
- Thinking cannot be turned off, forced tool use returns an error, and thinking blocks are tied to the model and conversation. On the Claude API and Google Cloud, computer_20251124 is not accepted.
- Published benchmark scores use Anthropic's harness and, unless noted, adaptive thinking at max effort. Terminal-Bench 4.0 is reported at xhigh effort. Reproduce before replacing Fable 5.1 or Opus 5.
- Biology and cybersecurity safeguards match the Fable 5.1 posture. Life-sciences and cyber verification programs are separate access paths, not the default API behavior.
Technical specs
- Inputs
- text, image
- Outputs
- text
- Capabilities
- long-running agentic coding, knowledge work, adaptive thinking, computer use, tool use, long context
- License
- Proprietary API
- Source
- https://www.anthropic.com/claude-opus-5-5
- Context tokens
- 1,000,000
- Hle With Tools
- 67.7
- Automationbench
- 40
- Cursorbench 4 0
- 57.8
- Knowledge Cutoff
- 2026-06
- Frontiercode V1 1
- 54.4
- Max output tokens
- 128,000
Benchmarks
- Gdpval Aa V2 1
- 1,846
- Terminal Bench 4 0
- 66.4
- Osworld 2 0 Partial
- 81.8
- Terminal Bench Science 0 1
- 58.7
Anthropic Claude family lineup
Current models
Previous versions
Related models
Claude Opus 5.5 FAQ
What is Claude Opus 5.5?
Claude Opus 5.5 is Anthropic's September 22, 2026 Opus-class model for long-running agentic coding and knowledge work, and the first model in the Claude 5.5 family. The Claude API ID is claude-opus-5-5, with a 1M-token context window, 128K synchronous max output, adaptive thinking that cannot be disabled, and default effort medium. Anthropic prices it at $4 input and $20 output per million tokens and says typical workloads cost about 40% less to run than Opus 5.
When does Claude Opus 5.5 fit best?
Long-running coding migrations and codebase audits
What should teams watch out for with Claude Opus 5.5?
Thinking cannot be turned off, forced tool use returns an error, and thinking blocks are tied to the model and conversation. On the Claude API and Google Cloud, computer_20251124 is not accepted.
Explore next
Models, tools, and comparisons that connect to this reference.