ChartFI - Coverage: leaderboard
ChartFI (2026) tests how faithfully and insightfully multimodal LLMs describe complex scientific charts, over 896 charts from peer-reviewed arXiv papers with author-grounded reference descriptions. This board is coverage: the F1 overlap between the insights in the generated description and those in the reference description, in percent; higher is better.
Metric: Coverage (%): F1 of the reference-description insights matched by generated insights (type-specific matching rules, threshold 0.7), times 100, descriptions of the 896 ChartFI-Bench scientific charts (arXiv papers accepted at peer-reviewed venues), temperature 0 and top-p 1; higher is better. Source: arxiv.org. Saturation forecast: Around 2030. 7 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 | 33.58 | #135 |
| 2 | Gemini 3 Flash | 31.45 | #175 |
| 3 | Qwen 3.5 Plus | 30.59 | #242 |
| 4 | Claude Sonnet 4.6 | 30.38 | #173 |
| 5 | Qwen 3.5 27B | 29.01 | #375 |
| 6 | Llama 4 Scout | 25.39 | #1818 |
| 7 | InternVL3.5-14B | 24.61 | #1504 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=chartfi-coverage · How It Works · Data refreshed daily, snapshot 2026-11-21.