Anthropic released Claude Opus 5 on Thursday. The model matches or beats its flagship Fable 5 on most coding and knowledge-work benchmarks, at half the token price. The launch makes Claude Opus 5 the fourth model Anthropic has shipped in under two months.
Fable 5 arrived in early June, followed by Sonnet 5 at the end of the month and now Opus 5 on July 24. Pricing holds steady at $5 per million input tokens and $25 per million output tokens, the same rate as its predecessor. Anthropic has not announced a replacement price tier for existing Opus users. The upgrade lands automatically for anyone already paying for the top model.
Claude Opus 5 tops Frontier-Bench and ARC-AGI 3
On Frontier-Bench, a coding evaluation that tests whether a model can build working software directly from engineering drawings, Opus 5 more than doubles its predecessor’s score and beats every competing model, including Fable 5. On ARC-AGI 3, which measures novel problem-solving, it scores roughly three times higher than the next-best model. On Zapier AutomationBench, a test of real business tasks completed end to end, Opus 5 passes more tasks than any rival even when running at its lowest effort setting.
Anthropic calls it the most aligned model yet
Anthropic says Opus 5 recorded the lowest rate of deceptive behavior of any model it has tested in its automated safety audit. Its predecessor set a new bar for honesty when it launched in late May, and Opus 5 pushes that further, though it still trails a system called Mythos 5 on offensive cybersecurity tasks. Anthropic says it deliberately avoided training the model on cyber work, and that the gains it did see there came as a side effect of broader capability improvements rather than targeted training.
Where to get Claude Opus 5
Opus 5 is live today as the new default model on Claude Max and the strongest option on Claude Pro. Developers can reach it through Anthropic’s API, Amazon Bedrock, Google Cloud, and Microsoft Foundry. A new Fast mode runs at two and a half times the default speed for double the base price.
Two API features ship alongside the model. One lets developers swap tools mid-conversation without breaking the prompt cache. The other automatically routes flagged requests to a different model instead of blocking them outright.
A faster release cadence
The rollout reflects a shift in how often Anthropic ships. The company has gone from one flagship release per quarter to four models in roughly eight weeks, each aimed at a different price-performance tier. Claude Opus 5 sits between the costlier Fable tier and the budget-focused Sonnet tier. It delivers near-frontier performance for everyday coding and knowledge work without the cybersecurity restrictions placed on Fable or the limited availability of Mythos. For teams already budgeting around Anthropic’s API pricing, the practical effect is a top-tier model available at the cost previously reserved for a mid-tier one.