Anthropic's Claude Opus 5 ships with dynamic tool changes and improved benchmarks

Anthropic has released Claude Opus 5, maintaining the price of Opus 4.8 while claiming near frontier intelligence and offering significant benchmark improvements. The release includes two beta features: the ability to dynamically change an agent's available tools mid-conversation without invalidating prompt caches, and automatic fallbacks for requests flagged by safety classifiers. The mid-conversation tool changes are particularly noteworthy, allowing for more flexible agent development by decoupling tool selection from the initial prompt. AI

IMPACT Enables more dynamic and flexible agent development by allowing mid-conversation tool adjustments, potentially improving user experience and reducing costs.

RANK_REASON Frontier-lab model release with system card and benchmark results. [lever_c_demoted from frontier_release: ic=1 ai=1.0]

Read on Towards AI →

• Amazon Bedrock
• Anthropic
• ARC AGI 3
• Claude Fable 5
• Claude Opus 5
• Frontier-Bench v0.1
• GDPval-AA
• Opus 4.8
• VentureBeat

• model release
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →