Anthropic releases Claude Opus 5, positioning it as a cost-efficient upgrade with near-frontier performance

Anthropic released Claude Opus 5, positioning it as a cost-efficient upgrade within its Opus model family. The company states Opus 5 is available today and delivers performance "close to the frontier intelligence of Claude Fable 5" at half the price.

On coding and knowledge work evaluations such as Frontier-Bench and GDPval-AA, Anthropic reports Opus 5 achieves state-of-the-art results. However, the model remains behind Mythos 5 on cybersecurity tasks.

Anthropic highlights Opus 5's efficiency improvements over its predecessor, Opus 4.8, noting that Opus 5 more than doubles Opus 4.8's performance on Frontier-Bench v0.1 at a lower cost per task. On CursorBench 3.2 at max effort, Anthropic reports Opus 5 performs within 0.5% of Fable 5's peak score while costing half as much per task.

For knowledge work and problem-solving tasks, Anthropic reports Opus 5 achieves three times the score of the next-best model on ARC-AGI 3, a 1.5× higher pass rate than the next-best model on Zapier AutomationBench at the same cost per task, and surpasses Fable 5's best result on OSWorld 2.0 at just over a third of the cost.

Anthropic also claims Opus 5 improves performance on scientific research tasks, including life sciences evaluations in structural biology, organic chemistry, and bioinformatics. The company reports Opus 5 scores 10.2 percentage points higher than Opus 4.8 on an internal organic chemistry benchmark and 7.7 percentage points higher on protein-related tasks.

Anthropic emphasizes Opus 5's enhanced ability to verify its work and iterate until success, citing examples where Opus 5 built a computer vision pipeline to extract geometry from raw pixels, fixed a previously missed edge case in an open-source package manager, and constructed a market data feed with its own test harness.