Anthropic just put out Opus 5, and the headline isn’t that it’s smarter. It’s that it’s half the price of what came before it, and in a bunch of tests it actually beats its own bigger, pricier sibling, Fable 5.

That’s the part worth sitting with. For a while the story in AI was simple: pay more, get more. Anthropic just broke that a little. Opus 5 scored higher on eight of thirteen benchmarks than Fable 5, especially on the stuff people actually use it for, coding and long agentic work, while costing about fifty percent less to run.

Anthropic says the bigger shift is under the hood. Opus 5 is built to check its own work and keep iterating instead of confidently handing you a half finished answer. In one test it wrote its own computer vision pipeline off an incomplete prompt, filling gaps nobody spelled out for it. That’s a different skill than being smart. That’s being careful.

There’s a dial too. You can tell it to spend more or less effort on a task, which is really a way of saying you decide how much thinking you’re paying for. For a company burning through Claude credits every month, that’s not a small thing.

It’s live now on the API, Bedrock, Google Cloud, Microsoft Foundry, claude.ai, Claude Code, and Cowork. No waitlist, no rollout schedule. It’s just there.

Anthropic is also leaning into safety with this one, saying the model needs far fewer safety interventions than Fable did. Whether that holds up under real use is something we’ll only know in a few months. But a cheaper, more careful model landing the same week competitors are racing toward bigger and pricier is worth noticing.

——

Follow: @Ali Demi
Book your free AI clarity call, NOW!
https://buff.ly/TpWy277

——

Sources:
https://www.anthropic.com/news/claude-opus-5
https://techcrunch.com/2026/07/24/anthropic-launches-opus-5/
https://www.axios.com/2026/07/24/anthropic-releases-new-model-opus-5

Repost this. Thanks.