GRAB-Lite: leaderboard
GRaph Analysis Benchmark: 500 task-balanced questions testing multimodal models on graph/figure analysis, estimating means, intercepts, correlations, and transforms across synthetic and realistic figures.
Metric: Overall Score. Source: grab-benchmark.github.io. Saturation forecast: Around January 2027. 38 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5 | 74 | #16 |
| 2 | GPT-5.5 | 71.8 | #31 |
| 3 | GPT-5.4 | 71 | #79 |
| 4 | Gemini 3.5 Flash | 63 | #60 |
| 5 | GPT-5.4 Mini | 62 | #212 |
| 6 | Claude Opus 4.8 | 60.6 | #37 |
| 7 | Gemini 3.1 Pro (Preview) | 59.4 | #58 |
| 8 | GPT-5.2 | 59 | #110 |
| 9 | Gemini 3 Flash | 58.2 | #99 |
| 10 | Claude Opus 4.7 | 58.2 | #49 |
| 11 | Gemini 3 Pro | 53.4 | #80 |
| 12 | Gemini 3.1 Flash Lite | 47.8 | #191 |
| 13 | Claude Sonnet 4.6 | 46.6 | #90 |
| 14 | GPT-5 | 46.2 | #98 |
| 15 | Claude Opus 4.6 | 45.6 | #63 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=grab-lite · How It Works · Data refreshed daily, snapshot 2026-10-20.