HELM TORR - Tablebench Data Analysis: leaderboard
The "Tablebench Data Analysis" subtask of HELM TORR. Stanford HELM's TORR project (Table-Oriented Reasoning and Retrieval): question answering and reasoning over tables and semi-structured records, including financial filings, scored on program and answer accuracy.
Metric: ROUGE-L. Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o (2024-11-20) | 32.99 | #379 |
| 2 | DeepSeek V3 | 31.49 | #325 |
| 3 | Claude 3.5 Sonnet (20241022) | 30.94 | #293 |
| 4 | Llama 3.1 405B Instruct | 29.91 | #464 |
| 5 | Llama 3.1 70B Instruct | 29.61 | #567 |
| 6 | Gemini 1.5 Pro (002) | 28.57 | #352 |
| 7 | Gemini 1.5 Flash (002) | 28.14 | #584 |
| 8 | Qwen 2 72B Instruct | 26.92 | #581 |
| 9 | GPT-4o Mini (2024-07-18) | 26.45 | #560 |
| 10 | Mixtral 8x22B Instruct | 23.71 | #735 |
| 11 | Mixtral 8x7B Instruct | 23.24 | #1171 |
| 12 | Claude 3.5 Haiku (20241022) | 23.12 | #591 |
| 13 | Mistral 7B Instruct (v0.3) | 21.91 | #1272 |
| 14 | Llama 3.1 8B Instruct | 15.84 | #1042 |
No result here: #4 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-torr-tablebench-data-analysis · How It Works · Data refreshed daily, snapshot 2026-10-24.