Microsoft: mai-image-2-efficient
2 benchmarks
Leaderboard position
#460
Overall rank (of 1121 models)
1169
Elo rating
2
Benchmarks
Top 41.0%
Percentile
In Microsoft (of 19 models)
Above: Google: Gemma 4 31B Instruct #459
Below: rednote-hulab: dots.ocr #461
Benchmark highlights
Nearest competitor from another company
The model from another company with the most similar benchmark scores.
appeared as nearest in 1 of 2 benchmarks
1/2
50.0%
1/2
50.0%
| Model | Votes | % of benchmarks |
|---|---|---|
| Black Forest Labs: flux.2-dev | 1 | 50.0% |
| Baidu: ernie-image | 1 | 50.0% |
Benchmarks
Predicted scores
Estimated from 25 similar models and the global Elo rating. These are projections, not measured results.
| Benchmark | Predicted | Confidence | Based on |
|---|---|---|---|
| MATH-500 | ~91.84 | 50% | — |
| AIME 2025 | ~76.42 | 50% | — |
| AIME 2026 | ~79.58 | 50% | — |
| HMMT Feb 2026 | ~71.11 | 50% | — |
| GPQA Diamond | ~65 | 50% | — |
| MMLU-Pro | ~68.94 | 50% | — |
| Humanity's Last Exam (no tools) | ~17.23 | 50% | — |
| IFEval | ~86.58 | 50% | — |
| IFBench | ~58.54 | 50% | — |
| HumanEval | ~82.84 | 50% | — |
| MBPP | ~89 | 50% | — |
| Terminal-Bench v2.1 | ~32.41 | 50% | — |
| MMLU | ~75.78 | 50% | — |
| SWE-bench Verified | ~49.06 | 50% | — |
| SWE-bench Pro | ~27.34 | 50% | — |
| SciCode | ~31.84 | 50% | — |
| AA-LCR | ~49.24 | 50% | — |
| Artificial Analysis Intelligence Index v5.0.1 | ~12 | 50% | — |
| Humanity's Last Exam (with tools) | ~23.33 | 50% | — |
| GSM8K | ~94.65 | 50% | — |
| -(CoderPPL) | ~-1.69 | 50% | — |
| LiveCodeBench v6 | ~63.12 | 50% | — |
| MMMU-Pro | ~58.86 | 50% | — |
| MATH | ~73.96 | 50% | — |
| BBH | ~72.31 | 50% | — |