IFMTBench - Single-Constraint: leaderboard
Metric: IF_Score (%): hard constraints checked by deterministic verifiers gate the item, soft constraints are scored by a gpt-oss-120b rubric judge, mean over items, on the 4,506 single-constraint items, translation requests into seven languages with instructions paraphrased in all seven, officially recommended decoding with non-thinking mode for open models; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 91.95 |
| 2 | Hy-MT2-30B-A3B | 91.94 |
| 3 | Gemma 4 31B | 84 |
| 4 | Gemma 4 26B A4B | 83.17 |
| 5 | Qwen 3.5 27B (Non-reasoning) | 82.48 |
| 6 | Qwen 3.6 35B A3B (Non-reasoning) | 77.95 |
| 7 | Hy-MT2-1.8B | 76.76 |
| 8 | Qwen 3.5 35B A3B (Non-reasoning) | 76.46 |
| 9 | Qwen 3.5 9B (Non-reasoning) | 75.59 |
| 10 | Gemma 4 E4B | 74.64 |
| 11 | Qwen 3.5 4B (Non-reasoning) | 72.58 |
| 12 | Gemma 4 E2B | 63.71 |
| 13 | Qwen 3.5 2B (Non-reasoning) | 54.77 |
| 14 | Qwen 3.5 0.8B (Non-reasoning) | 40.92 |
Interactive version: theaggregate.ai/benchmark?slug=ifmtbench-single-constraint · How It Works · Data refreshed daily, snapshot 2026-10-07.