Claude API quadruples its request size limit to 128 MB
Anthropic raised the Messages and Token Counting API cap from 32 MB to 128 MB on the first-party Claude API, easing a limit that pinched large files and computer-use agents.
Anthropic raised the maximum request size for the Messages API and the Token Counting API from 32 MB to 128 MB on 28 October, on the first-party Claude API. The Message Batches API stays higher still, at 256 MB, and the Files API at 500 MB. Sessions, Agents and Environments requests remain capped at 32 MB.
The old 32 MB ceiling was easy to hit once you accounted for encoding overhead: base64-encoding a file inflates it by roughly a third, so a 24 MB PDF was already close to the limit before you added instructions or conversation history. Exceeding the cap returns a 413 request_too_large error.
The increase applies only to the first-party API. Partner-operated platforms keep their own limits: Bedrock caps requests at 20 MB, and Claude Models on Google Cloud (Google managed) at 30 MB. Anthropic-operated platforms, Claude Platform on AWS, Claude Models on Google Cloud (Anthropic managed) and Microsoft Foundry, support Messages and Token Counting requests up to 32 MB.
Why it matters: anyone sending large PDFs, images or long conversation histories directly through the first-party API gets more headroom before hitting a 413, which matters most for computer-use and document-heavy agents that attach screenshots or files to every turn.
The caveat: if you’re on Bedrock or Google-managed Claude rather than the first-party API, none of this applies to you.