Anthropic released Claude Opus 5, its new flagship model, claiming performance near competitor Fable 5 while charging half the token price. The model scores 30.2 percent on ARC-AGI-3, a benchmark measuring novel problem-solving ability, nearly four times higher than GPT-5.6 Sol.

Opus 5 demonstrates top-tier results in coding and knowledge work tasks. The pricing advantage matters for enterprises running large-scale inference workloads. Cutting token costs in half while maintaining competitive performance directly impacts operating expenses for companies deploying AI models at scale.

The ARC-AGI-3 benchmark specifically tests generalization to novel problems without task-specific training. Opus 5's substantial lead here suggests stronger reasoning capabilities compared to the GPT baseline. This metric reflects real-world problem-solving more directly than tasks where models simply retrieve learned patterns.

Anthropic positions Opus 5 as a flagship offering in a competitive market where price and performance increasingly determine adoption. OpenAI's Fable 5 has dominated enterprise discussions, but equivalent capability at half the cost fundamentally shifts purchasing decisions. Organizations can either reduce spending with Opus 5 or deploy more compute resources for identical budgets.

The model fits Anthropic's strategy of targeting enterprise workloads where reliability and reasoning matter more than raw speed or scale. Businesses running large volumes of knowledge work, customer service automation, or coding assistance directly benefit from the cost efficiency without capability trade-offs.

Benchmark performance alone doesn't guarantee real-world deployment success. Implementation costs, API reliability, and integration friction determine whether enterprises switch models. However, matching performance at substantially lower cost removes a major barrier to adoption.

Anthropic's token pricing approach appeals to cost-conscious buyers without sacrificing capability. As AI markets mature, price-to-performance ratios become decisive.