Overview
- Anthropic released Claude Opus 5 on Friday, July 24, and deployed it across the Claude app and API as the default for Claude Max and the top model for Claude Pro.
- The company kept API pricing the same as the prior Opus release at $5 per million input tokens and $25 per million output tokens and added a Fast mode that generates about 2.5 times more output tokens per second for roughly twice the base price.
- Opus 5 introduces a low/medium/high effort toggle so users can trade reasoning depth for cost, and Anthropic’s published benchmarks say the model reaches near‑Fable performance on several coding and knowledge‑work tests, including within 0.5% of Fable 5 on CursorBench at max effort.
- Anthropic says Opus 5 applies much lighter automated safeguards than Fable 5, triggering about 85% fewer interventions, and it adds automatic fallback routing plus mid‑conversation tool switching so flagged requests return answers from other models instead of errors while still blocking exploit generation.
- The launch follows a rapid run of Anthropic releases that drew U.S. scrutiny in June, and it positions Opus 5 as a cost‑efficient, enterprise‑focused option that could lower token burn for routine work even as regulators and customers monitor safety across Anthropic’s model tiers.