Table Question Answering on MMTU
90.22AccuracyGPT-5-nano
Evaluation results
| Method | Links | |
|---|---|---|
| GPT-5-nanoWorkflow=Chain-of-Table (Tools), ST Reward=true, SR Reward=RG (TABROUGE)2026.01 | 90.22 | |
| GPT-5-nanoWorkflow=Tree-of-Table (Tools), ST Reward=true, SR Reward=RG (TABROUGE)2026.01 | 85.87 | |
| GPT-oss-20bWorkflow=Tree-of-Table (Tools), ST Reward=true, SR Reward=RG (TABROUGE)2026.01 | 81.52 | |
| GPT-4.1-nanoWorkflow=Chain-of-Table (Tools), ST Reward=true, SR Reward=RG (TABROUGE)2026.01 | 79.35 | |
| GPT-4.1-nanoWorkflow=Tree-of-Table (Tools), ST Reward=true, SR Reward=RG (TABROUGE)2026.01 | 78.26 | |
| GPT-oss-20bWorkflow=Chain-of-Table (Tools), ST Reward=true, SR Reward=RG (TABROUGE)2026.01 | 77.17 | |
| Ministral-3-8BWorkflow=Chain-of-Table (Tools), ST Reward=true, SR Reward=RG (TABROUGE)2026.01 | 54.35 | |
| QWEN3-8BWorkflow=Chain-of-Table (Tools), ST Reward=true, SR Reward=RG (TABROUGE)2026.01 | 52.17 | |
| Ministral-3-8BWorkflow=Tree-of-Table (Tools), ST Reward=true, SR Reward=RG (TABROUGE)2026.01 | 34.78 | |
| QWEN3-8BWorkflow=Tree-of-Table (Tools), ST Reward=true, SR Reward=RG (TABROUGE)2026.01 | 33.7 |