Announcement
Anthropic PBC announced on Friday, 24 July 2026, the launch of its new artificial‑intelligence model Claude Opus 5, positioned as a lower‑cost solution for workplace tasks amid rising price sensitivity and growing competition in China.
Pricing
Claude Opus 5 is priced at $5 per million input tokens and $25 per million output tokens, matching the cost structure of its predecessor Opus 4.8 while delivering substantially higher performance. This pricing represents a 50 % reduction relative to the company’s most advanced model, Fable 5.
Performance Benchmarks
- On the Frontier‑Bench v0.1 suite, Opus 5 outperformed all other evaluated models and delivered more than double the performance of Opus 4.8 at a lower cost per task.
- On CursorBench 3.2, Opus 5 achieved a score within 0.5 % of Fable 5’s peak performance when run at maximum effort, yet required only half the cost per task.
- For problem‑solving tasks measured by ARC‑AGI 3, Opus 5 scored three times higher than the next‑best model.
- In the Zapier AutomationBench, which assesses end‑to‑end business task completion, Opus 5’s pass rate was approximately 1.5 times that of the runner‑up model for the same cost per task.
- Scientific research benchmarks showed Opus 5 scoring 10.2 percentage points higher than Opus 4.8 on organic chemistry tasks and 7.7 percentage points higher on protein‑related tasks.
Safety and Risk
In internal safety testing, Opus 5 recorded an overall misaligned‑behavior score of 2.3, the lowest among recent Anthropic models, indicating improved alignment. The company emphasized that Opus 5 does not advance risky dual‑use capabilities and remains behind Mythos 5 in biology research and offensive cybersecurity domains.
Availability and Modes
Opus 5 is now the default model on Claude Max and the strongest offering on Claude Pro. Anthropic also provides a Fast mode that runs approximately 2.5 times faster than the default speed, priced at twice the base rate.