Anthropic Launches Claude Opus 5 AI, Halving Cost for High-End Performance

Anthropic has launched a new artificial intelligence model, Claude Opus 5, aiming to capture a larger share of the enterprise market by offering performance near its top-tier model at half the cost. The company announced on Friday that the new model is designed to handle a wide range of workplace tasks more affordably, a strategic move amid increasing price sensitivity from customers and intensifying competition.

A New Price-Performance Benchmark

Anthropic positioned Claude Opus 5 as a new standard for cost-effective AI, stating it delivers performance close to its most advanced model, Fable 5, for 50% less. The company expects the new model to become the default choice for many routine office and software development tasks.

Claude Opus 5 is priced at $5 per million input tokens and $25 per million output tokens. According to the company, this pricing matches its predecessor, Opus 4.8, while offering significantly improved capabilities. The model was made available across all of Anthropic's platforms upon its release.

Performance and Capabilities

Anthropic released several benchmarks to demonstrate the model's enhanced performance, particularly in complex, professional domains. The company highlighted its strong results on software engineering, problem-solving, and scientific research tasks.

Ad

• Software Engineering: On the Frontier-Bench v0.1 test, Opus 5 reportedly outperformed all other models and more than doubled the performance of Opus 4.8. On another test, CursorBench 3.2, it achieved a score within 0.5% of Fable 5 at half the cost.
• Problem-Solving: The model scored three times higher than the next-best model on the ARC-AGI 3 benchmark.
• Business Automation: On the Zapier AutomationBench, which tests end-to-end task completion, Opus 5's pass rate was approximately 1.5 times that of the next-best model for the same cost.

Market Context and Safety

The launch reflects a broader trend in the AI industry, where providers are competing not just on capability but also on price as enterprise adoption grows. The source material notes that this move is a response to both customer cost concerns and growing competition, including from developers in China.

In its safety evaluations, Anthropic reported that Opus 5 scored a 2.3 on overall misaligned behavior, the lowest among its recent models. The company also stated the model does not advance risky, dual-use capabilities and remains less proficient than its specialized Mythos 5 model in sensitive areas like biology research and offensive cybersecurity.