September 23, 2026
Working with Opus 5.5 got cheaper: API is 20% below Opus 5
On September 22, Anthropic cut the API price of Opus 5.5 by 20% versus Opus 5: $4 instead of $5 per million input tokens and $20 instead of $25 per million output tokens. The model has a 1 million-token context window and a maximum response of 128 thousand.
Addy Osmani
@addyosmani
Opus 5.5 has become my new everyday model. On most tasks, it performs at Fable's level, but costs less and runs faster than Opus 5. And Pro, Max, and Team users have just received a limit reset they can save for later. Demo:
· 27K views
Opus 5.5 is available in the Claude API, AWS Bedrock, Google Cloud, and Microsoft Foundry.
In Claude Code. Previously, sessions were launched on Opus 5. The model can now be enabled with `claude --model claude-opus-5-5` or through `/model` in an already running agent.
In the API. The ID `claude-opus-5` changes to `claude-opus-5-5`. When migrating, remove `thinking`, select `effort`, and replace `tool_choice: any` or `tool` with `auto` using strict tool use.
Anthropic says Opus 5.5 costs 40% less on typical tasks and responds more than 30% faster than Opus 5. Cache reads cost $0,20 per million tokens instead of $0,50.
In Fast mode, the model runs up to 2,5 times faster in Claude Code and Claude Platform. The mode costs $8 per million input tokens and $40 per million output tokens.
Anthropic reports 66,4% on Terminal-Bench 4.0 versus 52,3% for Opus 5, and 57,8% versus 46,6% on CursorBench 4.0. Stripe engineer Christian Rivera said the model rebased 40 related pull requests, and all of them passed CI the next day.
For Pro, Max, Team, and Enterprise, Anthropic increased five-hour limits and issued a limit reset that can be saved for later.
In the Batch API with the beta header, Opus 5.5 can generate up to 300 thousand tokens per response.
