Anthropic released Claude Opus 5 on July 24, 2026, delivering a dramatic leap in the Opus tier — approaching the frontier intelligence of Claude Fable 5 at half the cost.
Performance at a fraction of the cost
Opus 5 tops Frontier-Bench v0.1, surpassing all other models while more than doubling Opus 4.8's performance at a lower cost per task. On CursorBench 3.2, at max effort, it performs within 0.5% of Fable 5's peak score at half the cost per task.
On Zapier AutomationBench, Opus 5's pass rate is roughly 1.5× the next-best model for the same cost. Even at its lowest effort setting, it passes more tasks than any other model. On OSWorld 2.0, a computer-use benchmark, it outperforms every model at any given cost, surpassing Fable 5's best result at just over a third of the cost.
Exceptional agency and reasoning
Opus 5 demonstrates remarkable autonomy. In one Frontier-Bench task, given a drawing of a machine part with no direct way to view it, the model wrote its own computer vision pipeline to extract geometry from raw pixels and reconstructed a full 3D FreeCAD model — something no competing model could do after five attempts.
Given a real bug in a popular open-source package manager, Opus 5 found the root cause and fixed an edge case that the community's patch had missed — while a competing model fixed only the surface symptom.
Scientific and professional gains
On life sciences evaluations, Opus 5 shows meaningful improvements across structural biology, organic chemistry, and bioinformatics. On organic chemistry tasks like inferring molecular structures from spectroscopy data, it scores 10.2 percentage points higher than Opus 4.8. On protein sequence variation tasks, it improves by 7.7 percentage points.
On ARC-AGI 3, which tests novel problem-solving, Opus 5 scores three times as high as the next-best model.
Safety and alignment
Opus 5 scored 2.3 on Anthropic's overall misaligned behavior scale — the lowest of any recent model, making it the company's most-aligned model to date. It exhibits the lowest rates of deceptive behavior and is the least susceptible to being tricked into misuse.
Critically, the model does not advance the frontier in risky dual-use capabilities. It remains behind Mythos 5 in both biology research and offensive cybersecurity, though it shows substantial improvement at finding vulnerabilities.
Pricing and availability
Claude Opus 5 is priced at $5 per million input tokens and $25 per million output tokens — the same as Opus 4.8. It is the new default on Claude Max and the strongest model on Claude Pro, available on all platforms and the Claude API.
A Fast mode option runs at roughly 2.5× the default speed at twice the base price. Anthropic also released mid-conversation tool changes on the Claude Platform and automatic fallbacks on the API as new beta features.
