Benchmarks · instruction
IFBench
IFBench: precise instruction-following with verifiable constraints; accuracy.
98 models · 157 results · data as of 2026-10-29 · artificialanalysis.ai ↗
Leaderboard
Grok 4.3 medium
More from this benchmark
3 published measuresScore vs any published measure3 measures: sub-scores, splits, counts
Breakdown
darker is better within each column · click a column to sort| Model | |||
|---|---|---|---|
| Grok 4.3 medium | 61,328 | 33,597 | 754,087 |
| MiniMax M3 | 66,430 | 90,906 | 3,142,033 |
| Qwen 3.7 Max | 26,842 | 101,220 | 1,589,710 |
| MiMo V2.5 Pro | 31,029 | 144,544 | 1,069,409 |
| DeepSeek V4 Flash 0420 max | 48,106 | 60,958 | 3,984,764 |
| Qwen 3.5 397B-A17B | 24,589 | 69,129 | 2,639,979 |
| Qwen 3.7 Plus | 26,842 | 95,129 | 1,855,904 |
| GPT-5.2 Codex xhigh | 24,562 | 62,229 | 3,671,194 |
| DeepSeek V4 Pro 0424 max | 48,106 | 71,728 | 3,188,954 |
| Gemini 3.5 Flash high | 24,101 | 121,451 | 1,840,551 |
| GLM 5.1 | 25,200 | 100,819 | 1,645,081 |
| Kimi K2.6 | 24,589 | 94,111 | 2,408,598 |
| Muse Spark | 41,783 | 96,617 | 1,163,132 |
| GPT-5.4 Nano xhigh | 24,589 | 125,137 | 3,538,153 |
| GPT-5.5 xhigh | 32,161 | 152,070 | 913,358 |
| MiniMax M2.7 | 24,589 | 109,381 | 4,164,307 |
| Qwen 3.5 122B-A10B | 24,589 | 360,072 | 3,137,258 |
| Gemma 4 31B | 24,572 | 66,945 | 872,203 |
| Qwen 3.5 27B | 24,589 | 311,089 | 3,414,995 |
| GPT-5 Mini high | 24,896 | 146,784 | 2,453,504 |
| GPT-5.2 xhigh | 24,589 | 143,899 | 2,159,888 |
| GPT-5.3 Codex xhigh | 24,589 | 86,822 | 1,031,956 |
| Qwen 3.6 Plus | 24,589 | 134,710 | 2,413,509 |
| GPT-5 Codex high | 24,589 | 88,887 | 948,032 |
| GPT-5.4 xhigh | 24,589 | 132,732 | 2,229,703 |
All results
157 rows| # | Model | Effort | IFBench |
|---|---|---|---|
| 1 | Grok 4.3 | medium | 83.3% |
| 2 | MiniMax M3 | 82.9% | |
| 3 | Grok 4.3 | high | 81.3% |
| 4 | Grok 4.3 | low | 81.0% |
| 5 | Qwen 3.7 Max | max | 80.5% |
| 6 | MiMo V2.5 Pro | 79.9% | |
| 7 | DeepSeek V4 Flash 0420 | max | 79.2% |
| 8 | Qwen 3.5 397B-A17B | 78.8% | |
| 9 | Qwen 3.7 Plus | 78.0% | |
| 10 | GPT-5.2 Codex | xhigh | 77.6% |
| 11 | DeepSeek V4 Pro 0424 | max | 76.5% |
| 12 | Gemini 3.5 Flash | high | 76.3% |
| 13 | GLM 5.1 | 76.3% | |
| 14 | Kimi K2.6 | 76.0% | |
| 15 | Muse Spark | 75.9% | |
| 16 | GPT-5.4 Nano | xhigh | 75.9% |
| 17 | GPT-5.5 | xhigh | 75.9% |
| 18 | MiniMax M2.7 | 75.7% | |
| 19 | Qwen 3.5 122B-A10B | 75.7% | |
| 20 | Gemma 4 31B | 75.6% | |
| 21 | Qwen 3.5 27B | 75.6% | |
| 22 | GPT-5 Mini | high | 75.4% |
| 23 | GPT-5.2 | xhigh | 75.4% |
| 24 | GPT-5.3 Codex | xhigh | 75.4% |
| 25 | Qwen 3.6 Plus | 75.2% | |
| 26 | Gemini 3.5 Flash | medium | 74.6% |
| 27 | GPT-5 Codex | high | 74.1% |
| 28 | GPT-5.4 | xhigh | 73.9% |
| 29 | Gemma 4 12B | 73.5% | |
| 30 | DeepSeek V4 Flash 0420 | high | 73.5% |