Anthropic launches Claude Opus 5.5 for agentic coding

Anthropic launched Claude Opus 5.5 on September 22, 2026 as the first model in its Claude 5.5 family. The company says it performs at the level of Claude Fable 5.1 on most work and costs about 40% less to run than Opus 5. Reuters reported the launch and attributed the Fable comparison and cheaper-to-run figure to Anthropic; it did not independently verify performance or cost.

For teams that price long agent runs on tokens, cache, and turns, the list prices are the practical change. Input and output tokens are $4 and $20 per million — 20% below Opus 5. Cache reads, which Anthropic says make up most of the cost of agentic and coding work, are $0.20 per million tokens, 60% below Opus 5. Cache writes are $5 per million. Anthropic also says Opus 5.5 generates output more than 30% faster than Opus 5. A separate fast mode in Claude Code and the Claude Platform is priced at $8 / $40 per million input/output tokens, with up to 2.5× speed.

The Claude API model ID is claude-opus-5-5. Specs are a 1 million token context window, 128K max output, and adaptive thinking that is always on; default effort is medium. The same ID is used on Google Cloud, Microsoft Foundry, and Claude Platform on AWS. Amazon Bedrock’s ID is anthropic.claude-opus-5-5. Four breaking changes hit code already on Opus 5: thinking cannot be disabled, forced tool use returns an error, thinking blocks are tied to the model that produced them, and the earlier computer_20251124 computer-use tool is not accepted on the Claude API and Google Cloud.

AWS posted same-day availability on Amazon Bedrock and Claude Platform on AWS. On Bedrock, Opus 5.5 runs through US, EU, AU, JP, and Global CRIS inference profiles (us., eu., au., jp., global.). In the Bedrock console, choose Test, then Playground, and select Claude Opus 5.5.

Anthropic says Frontier Design and METR tested the model before release, and that it is its strongest result to date on an internal automated behavioral audit. Reuters repeated those company statements; it did not independently verify the evals. Claude Sonnet 5.5 and Claude Haiku 5.5 are due in the coming weeks.

Try claude-opus-5-5 on the Claude API, or open the Amazon Bedrock Playground and select Claude Opus 5.5. Then read the model overview for effort, thinking, and the Opus 5 breaking changes before you migrate agents.

Sources