What happened
Anthropic released Claude Opus 4.5, a new large language model positioned as the company's best model to date. The model scored higher than any human on a two-hour intensive engineering task, demonstrating superior performance in coding, agentic tasks, and computer use. Claude Opus 4.5 is now available across every major cloud platform for the first time.
Why it matters
- The model's ability to pause and think before acting reduces wasted steps and improves results in enterprise workflows and everyday tasks like spreadsheets.
- Claude Opus 4.5 achieves significant accuracy improvements on tool use evaluations, with performance jumping from 79.5% to 88.1% when Tool Search is enabled, while reducing token usage by 85%.
- The model's enhanced front-end capabilities and vision performance make it substantially better at computer use and visual tasks.
Key facts
- 1. Claude Opus 4.5 scored higher than any human has ever scored on Anthropic's two-hour engineering assignment.
- 2. The model achieved an 85% reduction in token usage while maintaining access to full tool libraries.
- 3. On MCP evaluations with Tool Search enabled, Opus 4.5 improved accuracy from 79.5% to 88.1%.
- 4. Claude Opus 4.5 is available on every major cloud platform.
- 5. The model demonstrates superior performance in coding, agentic tasks, front-end work, and vision capabilities compared to previous versions.
Key Voices
- Anthropic engineers stated the model "gets it" and solves bugs that previous versions could not find, with colleagues reporting longer intervals between necessary interventions.
- Anthropic highlighted that Claude Opus 4.5 knows when to think before acting, enabling more efficient and reliable performance in complex workflows.
