Anthropic has released Claude Opus 5, and it is already available across the company's products. According to TestingCatalog, the model is live on all of Anthropic's platforms, where it becomes the new default for Claude Max subscribers and the most capable option offered on Claude Pro. Developers can call it through the API under the name claude-opus-5.
The early benchmark story is a favorable one. The-decoder.com reports that Opus 5 pulls ahead of both Fable 5 and GPT-5.6 Sol on a benchmark specifically designed to measure real intelligence rather than memorized test-taking. MLQ.ai reports that the model tops an AI benchmark index while costing half as much as Fable 5 — a combination that matters more than raw scores for anyone running these systems at scale. Chosun Ilbo frames the release in similar terms, describing Opus 5 as cheaper and faster.
The picture is not a clean sweep. Anthropic says Opus 5 beats Fable 5 on agentic search — the kind of multi-step work where a model goes and finds information rather than answering from memory. But as reporting carried by MSN notes, the new model still trails Fable 5 in some categories, including answering legal questions and performing multidisciplinary reasoning without the help of additional tools.
That mix is worth paying attention to. The frontier AI race has largely been narrated as a scoreboard, with each release claiming the crown. What the Opus 5 coverage suggests is a more segmented reality: different models lead in different task categories, and price is becoming as much of a competitive weapon as capability.
Why it matters: when a top-scoring model costs half what its closest rival does, advanced AI stops being a premium line item and starts becoming infrastructure that far more companies can actually afford to build on.