A zoom tool pushes chart-reading accuracy to 73%
Anthropic shared numbers from the Chartography benchmark, a set of 100 questions on dense charts built from real data. Give the model a zoom tool and Fable 5 jumped from 29% to 73% accuracy, while Sonnet 5 went from 13% to 44%.
Shrinking large images loses fine detail. The tool lets Claude request a specific region and get that slice back from the original at full resolution. The updated recipe is in Claude Cookbooks on GitHub.
Related stories
- Anthropic points developers to fresh examples in Claude Cookbooks
- Opus 5.5 aced a test suite at 3.4x GPT-6 Sol's cost
- Opus 5.5 finds new bugs for CodeRabbit and misses old ones
- Fable 5.1 refuses the knife but heats a gas can anyway
- Andon Labs puts GPT-6 Astra ahead of Claude Fable 5.1
- Claude Opus 5 tops Sierra's agent-building test at 23.9%
Comments
No comments yet. Be the first.
Join the conversation
Sign in with Google to leave a comment. Your name and avatar come from your Google profile, and the comment appears after moderation.