Anthropic’s Claude Opus 5 reportedly matches Opus 4.8 pricing and lifts key agent benchmarks

歸藏的AI工具箱 · wechat · 2026-07-25

Anthropic’s Claude Opus 5 is reportedly priced the same as Opus 4.8, while posting broad benchmark gains.

In agentic coding terminal tasks, Opus 5 scores 43.3%, ahead of Fable5’s 33.7% and GPT-5.6Sol’s 34.4%.
On computer-use tasks it reaches 70.6%, and business workflow tasks rise from 17.0% on Opus 4.8 to 26.0%.
On novel problem solving, it scores 30.2%, far above GPT-5.6Sol’s 7.8% and Opus 4.8’s 1.5%.
The post also jokes that the “agentic coding” score of 53.4% is labeled as slightly below Fable5’s 53.5%.

The poster concludes that Opus 5 finally looks usable and may make Claude Max less attractive on value.

More from Models