Chartography benchmark shows zoom tool lifts Fable 5 from 29% to 73%
echen · x · 2026-07-28
HelloSurgeAI’s Chartography benchmark tests whether models can read professional charts. In the quoted result, adding a zoom tool lifts Fable 5 from 29% to 73% accuracy and Sonnet 5 from 13% to 44%, suggesting chart-reading is highly sensitive to interface support rather than raw model ability alone.
Related event: Claude's Zoom Tool Significantly Boosts Chart Recognition Accuracy(3 posts)→
More from Research
- Yale PhD student open-sources his paper figure scripts, packaged as a Skill for Claude Code and Cursor — burny_tech · 2026-09-23
- AI models now match superforecasters on ForecastBench; rematch set for October — burny_tech · 2026-09-23
- Dev uses Opus 5.5 with Lean to formally verify Claude Agent SDK, yielding 16 bug-fix PRs — bcherny · 2026-09-23
- Mathematicians, not just LLMs, made AI's math breakthroughs possible, scholars argue — tak3sh8 · 2026-09-23
- AI-enabled drug discovery cuts discovery time by 15-80%, McKinsey research finds — menhguin · 2026-09-23
- Gemini training details dissected: groupwise reward redistribution to fight reward hacking — nrehiew_ · 2026-09-23