Microsoft: phi-3.5-mini-instruct
1 benchmarks
Leaderboard position
#722
Overall rank (of 1122 models)
619
Elo rating
1
Benchmarks
Top 64.3%
Percentile
In Microsoft (of 19 models)
Above: Sber: gigachat3.5-432b-a28b-base #721
Below: Meta Llama: Llama 3.3 70B Instruct #723
Benchmark highlights
Best benchmark
GSM8K
86.2
Worst benchmark
GSM8K
86.2
Nearest competitor from another company
The model from another company with the most similar benchmark scores.
appeared as nearest in 1 of 1 benchmarks
1/1
100.0%
Benchmarks
Predicted scores
Estimated from 25 similar models and the global Elo rating. These are projections, not measured results.
| Benchmark | Predicted | Confidence | Based on |
|---|---|---|---|
| Artificial Analysis Intelligence Index v5.0.1 | ~12 | 50% | — |
| MATH-500 | ~82.33 | 50% | — |
| AIME 2025 | ~58.44 | 50% | — |
| AIME 2026 | ~68.99 | 50% | — |
| HMMT Feb 2026 | ~60.46 | 50% | — |
| GPQA Diamond | ~52.94 | 50% | — |
| MMLU-Pro | ~55.49 | 50% | — |
| Humanity's Last Exam (no tools) | ~7.42 | 50% | — |
| IFEval | ~79.08 | 50% | — |
| IFBench | ~49.47 | 50% | — |
| HumanEval | ~66.92 | 50% | — |
| MBPP | ~68.14 | 50% | — |
| Terminal-Bench v2.1 | ~13.05 | 50% | — |
| MMLU | ~61.65 | 50% | — |
| SWE-bench Verified | ~31.95 | 50% | — |
| SWE-bench Pro | ~11.35 | 50% | — |
| SciCode | ~23.18 | 50% | — |
| AA-LCR | ~38.46 | 50% | — |
| Humanity's Last Exam (with tools) | ~11.9 | 50% | — |
| -(CoderPPL) | ~-2.37 | 50% | — |
| LiveCodeBench v6 | ~50.59 | 50% | — |
| MMMU-Pro | ~48.53 | 50% | — |
| MATH | ~50.95 | 50% | — |
| BBH | ~60.15 | 50% | — |
| ARC-Easy | ~77.67 | 50% | — |