Anthropic Releases Claude Opus 5; Gradual Rollout in GitHub Copilot for Eligible Plans | Model Release
Anthropic released Claude Opus 5 on July 24, 2026 as a new AI model for complex agentic coding and enterprise work, available via the API under model ID claude…
Published on MyPrivateClaw
Jul 25, 2026, 3:02 AM UTC
Coverage date
Jul 24, 2026
Last updated
Jul 25, 2026, 6:50 AM UTC
News summary
Claude Opus 5 has a 1M token context window (default and maximum) and supports up to 128k max output tokens, with an optional beta header output 300k 2026 03 24 enabling up to 300k output tokens. Claude Opus 5 is priced at $5 per million input tokens and $25 per million output tokens on the Claude API, unchanged from Claude Opus 4.8, while Claude Opus 5 Fast mode runs at approximately 2.5x the default speed and is priced at $10 per million input tokens and $50 per million output tokens; it is available only on the Claude API and not yet on Amazon Bedrock, Google Cloud, or Microsoft Foundry. Thinking is on by default in Claude Opus 5; disabling thinking requires the effort parameter to be set to high or below, and setting thinking disabled with xhigh or max effort returns a 400 error—a breaking change from Claude Opus 4.8. Copilot Enterprise and Copilot Business plan administrators must…