Claude Opus 5 is out. If you pay for Claude, you have it.

Anthropic released Claude Opus 5 on 24 July. For a founder, the story is not that it is smarter than what came before — Anthropic openly says its top model, Fable 5, is still the smartest. The story is the price. Opus 5 costs $5 per million input tokens and $25 per million output tokens, unchanged from Opus 4.8, while scoring far higher on several benchmarks. If you pay for Claude, you get more for the same money.

It is now the default model on Claude Max and the strongest model on Claude Pro, so subscribers do not need to do anything to start using it. Build on the API instead and the model id is claude-opus-5. A new adjustable effort setting lets you trade capability for speed and lower token use per request — the lever that actually changes your bill.

What the numbers say, and don't

On Frontier-Bench v0.1, an agentic coding test, Anthropic reports Opus 5 at 43.3% against Opus 4.8's 18.7% and Fable 5's 33.7%, at a lower cost per task. On the OSWorld computer-use benchmark it says Opus 5 beats Fable 5's best result at just over a third of the cost. Zapier's chief executive Wade Foster said the model topped his company's automation leaderboard “without spending more tokens than prior Claude models.”

The caveats are real and worth reading. Anthropic still points to Fable 5 for long, multi-day autonomous jobs where a model must stay coherent over hours. And Opus 5 does not win everywhere: per VentureBeat, it trails one competing model on cybersecurity and biology work, and an OpenAI-family model still leads on one coding benchmark.

What to do

• On Claude Pro or Max, there is nothing to buy and nothing to switch — you already have it. Your bill does not change.
• If you build on the API, the token price matches Opus 4.8, so the saving comes from the effort dial: run routine work at a lower setting and reserve the high setting for hard tasks. Test it on your own workload before you assume a number.
• For long-running autonomous agents, do not treat cheaper as better by default — Anthropic itself recommends comparing Opus 5 and Fable 5 on one bounded task and one long-horizon job.