Anthropic claims its new Claude Opus 5 delivers near-Fable 5 performance at half the token price

What happened

Anthropic launched Claude Opus 5, a new flagship large language model that edges close to Fable 5’s performance while cutting token costs in half. On ARC-AGI-3, a benchmark measuring novel problem-solving, Opus 5 scored 30.2 percent, nearly four times better than GPT-5.6 Sol, a strong competitor from other AI labs. The model also excels at coding and knowledge work tasks, positioning itself as a top-tier performer at a lower price point.

Why it matters

Claude Opus 5’s mix of high performance and lower token pricing challenges the current cost structures for advanced AI services. This puts pressure on existing leaders to justify their price points, forcing a reckoning on model efficiency versus user-cost trade-offs. For businesses that deploy large language models at scale, especially in coding or research workflows, this could reduce operational expenses meaningfully without sacrificing capability. Investors and operators will want to watch how this shifts the competitive dynamics around pricing and performance in the AI models market.

What to watch next

The next test will be how Anthropic’s model performs in real-world applications beyond benchmark scores and how service providers pass these cost savings to customers. Also critical is whether Fable 5 or rivals respond with improved price-performance ratios or new features that redefine the value equation. Watch for announcements on API pricing, commercial integrations, and how developers adopt Opus 5 for cost-sensitive projects. Ultimately, the winner will be whoever balances price, performance, and reliability best in deployment at scale.

AI Quick Briefs Editorial Desk