Generated

Anthropic releases Claude Opus 5, a cost-effective AI model that leads the ARC-AGI-3 benchmark with superior reasoning capabilities.

Anthropic has launched Claude Opus 5, its newest artificial intelligence model designed for daily use and complex coding tasks. The model achieved a significant milestone by scoring 30.2 percent on the ARC-AGI-3 benchmark, far surpassing the previous record of 7.8 percent set by OpenAI's GPT-5.6 Sol. Researchers credit this lead to stronger logical reasoning, which allows the model to autonomously explore and plan across unfamiliar environments. While Claude Opus 5 offers high performance, the administration announced its release as a more cost-effective alternative to the Fable 5 model, priced at $5 per million input tokens and $25 per million output tokens. It is positioned as a high-quality offering for enterprises looking for value, as it lags behind the Mythos 5 model in specialized cybersecurity and biology research. The model is now available in GitHub Copilot for Pro+ and Enterprise users, featuring enhanced safeguards for high-harm cyber content. Anthropic continues to collaborate with government agencies to ensure the model's reliability and safety before deployment.

Sources