Other · model
Phi-4-reasoning benchmark scores
As of 2026-10-29, Phi-4-reasoning (Other) has measured scores on 12 benchmarks and estimated scores on 32 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 57.5% | measured |
| AA-HLE | 3.8% | measured |
| AIME26 | 64.9% | estimated ± 4.6 pp, medium confidence |
| CritPt | 0.0% | measured |
| GDPval-AA | 0.0% | estimated ± 10.9 pp, low confidence |
| GPQA Diamond | 66.1% | estimated ± 1.2 pp, high confidence |
| HLE | 6.9% | estimated ± 6.9 pp, medium confidence |
| IFBench | 53.6% | estimated ± 6.6 pp, low confidence |
| AA-LCR | 0.0% | measured |
| IFEval | 83.4% | measured |
| LongBench v2 | 47.1% | estimated ± 5.8 pp, low confidence |
| τ²-bench results | 0.0% | measured |
| BioMysteryBench (human-difficult) | 0.0% | estimated ± 6.8 pp, low confidence |
| SWE-bench Verified | 54.7% | estimated ± 9.3 pp, low confidence |
| AA Agentic Index | 0.0% | estimated ± 12.3 pp, low confidence |
| MMLU-Pro | 74.3% | measured |
| AA-IFBench | 23.5% | measured |
| AIME 2024 | 75.3% | measured |
| Artificial Analysis Intelligence Index | 10.3% | estimated ± 5.8 pp, medium confidence |
| BBH | 92.4% | estimated ± 2.8 pp, medium confidence |
| GPQA | 65.8% | measured |
| MMLU | 86.1% | estimated ± 2.2 pp, high confidence |
| AA-Omniscience Accuracy | 20.9% | estimated ± 7.8 pp, medium confidence |
| MATH-500 | 96.8% | estimated ± 1.2 pp, high confidence |
| MMMLU | 83.3% | estimated ± 0.7 pp, medium confidence |
| AA MMLU-Pro | 84.3% | estimated ± 1.0 pp, low confidence |
| AIME 2025 | 62.9% | measured |
| ARC-AGI-1 | 74.7% | estimated ± 7.8 pp, low confidence |
| ARC-AGI-2 | 30.7% | estimated ± 14.6 pp, low confidence |
| Vals GPQA Diamond | 43.8% | estimated ± 7.1 pp, low confidence |
| Vals MMLU-Pro | 80.1% | estimated ± 2.3 pp, low confidence |
| HLE w/o tools | 11.6% | estimated ± 5.2 pp, low confidence |
| ARC-AGI-3 | 1.4% | estimated ± 7.7 pp, low confidence |
| C-Eval | 90.8% | estimated ± 1.1 pp, low confidence |
| HMMT Feb 2025 | 29.9% | estimated ± 3.1 pp, medium confidence |
| MMLU-Redux | 86.9% | estimated ± 0.6 pp, high confidence |
| SuperGPQA | 46.3% | estimated ± 8.2 pp, low confidence |
| HealthBench Hard | 0.0% | estimated ± 8.6 pp, low confidence |
| MMLU-Pro (Arcee) | 77.1% | estimated ± 4.0 pp, low confidence |
| BioMysteryBench (human-solvable) | 81.7% | estimated ± 0.6 pp, low confidence |
| STEMBench | 2.4% | estimated ± 0.2 pp, medium confidence |
| SimpleQA | 17.2% | estimated ± 7.8 pp, medium confidence |
| LiveCodeBench | 53.8% | measured |
| MRCRv2 | 22.8% | estimated ± 11.3 pp, low confidence |