MMTU - Table QA: leaderboard
The "Table QA" subtask of MMTU. Massive Multi-Task Table Understanding benchmark with real-world table reasoning tasks including joins, transformations, cleaning, and question answering.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 26 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 89.48 | #98 |
| 2 | GPT-5 Mini | 87.7 | #186 |
| 3 | Gemini 2.5 Pro | 87.61 | #152 |
| 4 | GPT-OSS-120B | 85.96 | #341 |
| 5 | Grok 3 Mini | 85.9 | #281 |
| 6 | DeepSeek R1 0528 | 85.71 | #222 |
| 7 | O3 | 85.11 | #124 |
| 8 | Gemini 2.5 Flash | 84.89 | #245 |
| 9 | Qwen 3 235B A22B 2507 (Thinking) | 83.12 | #260 (Qwen 3 235B A22B 2507) |
| 10 | GPT-OSS-20B | 82.83 | #518 |
| 11 | Qwen 3 32B | 82.51 | #443 |
| 12 | Qwen 3 8B | 80.76 | #674 |
| 13 | Llama 4 Maverick Instruct FP8 | 79.81 | #414 |
| 14 | Llama 4 Scout Instruct | 73.98 | #539 |
| 15 | GPT-5 Chat | 72.85 | #241 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mmtu-table-qa · How It Works · Data refreshed daily, snapshot 2026-10-20.