Product Launches and Community Projects

Launch: Claude Opus 5.5 (Anthropic)

Posted by Applied AI Commons team (community team) on

Anthropic's new top model for agentic coding and knowledge work, at lower cost

First model in Anthropic's Claude 5.5 family, aimed at agentic coding, computer use and knowledge work. It uses fewer tokens per task than Opus 5, is priced 20% lower ($4/$20 per M tokens) and is available on the API, AWS, Google Cloud and Azure. Anthropic notes that benchmark margins overstate real-world differences, and most cyber tasks are routed to Opus 4.8.

Released 2026-09-22 by Anthropic. Listed by the community team; we are not affiliated.

If you've moved long-running coding or research agents to Opus 5.5, what changed in cost per task and in reliability?

Link: https://www.anthropic.com/claude-opus-5-5

3 replies

  1. Applied AI Commons team (starter reply from the community team),

    The pricing is $4 input and $20 output per million tokens, 20% lower than Opus 5, and Anthropic also says it uses fewer tokens per task. Both matter for long agent runs.

  2. Applied AI Commons team (starter reply from the community team),

    Anthropic itself notes that benchmark margins overstate real-world differences, which is a good reason to test on your own tasks before switching.

  3. Applied AI Commons team (starter reply from the community team),

    Availability on AWS, Google Cloud and Azure as well as the direct API helps teams that need to keep spend and data inside an existing cloud agreement.

More in Product Launches and Community Projects