AI Pulse · AI Trends Pulse

Anthropic shipped Claude Opus 5 this morning with a feature that hasn’t shown up in a frontier model before: a dial that lets you choose how hard the model works on each prompt. You pick low, medium, or high effort. Low costs less and runs faster. High costs more and thinks longer. You decide per task.

The pricing sits around half what the previous Opus tier charged, and Opus 5 is now the default for anyone on Claude Max. This is Anthropic’s fourth model release in less than two months, which is a pace we haven’t seen from them before.

What this means if you’re paying for AI today

If you’re running Claude in production, you now have a cost lever you can pull without switching models. A bulk email rewrite can run on low effort. A contract review or a technical spec can run on high. That’s useful if you’re trying to keep a monthly bill predictable while still getting access to the strongest reasoning when you need it.

The effort toggle also matters because it’s a design pattern other providers will likely copy. We’re moving past the era where you pick a model and accept whatever it costs per token. Now you’re managing a portfolio of capabilities and budgets inside a single interface. That’s the kind of workflow decision we build into the Omni Command Centre, so teams can route tasks intelligently without someone manually choosing a tier every time.

Four releases in eight weeks suggests Anthropic is trying to hold ground against OpenAI and Google, both of which have been shipping faster than usual. For buyers, that’s good. It means more options and downward pressure on price. For operators, it means your AI stack is going to need more frequent review than it did six months ago.

You can read the original announcement and the coverage from Fortune and Anthropic directly for technical benchmarks and API details.