# MMLU-ProX → SWE Multilingual

Calibration · https://benchgap.net/calibration/mmluprox/current/to/swe-bench-multilingual/current · data retrieved 2026-11-18

SWE Multilingual is estimated from MMLU-ProX with a inverse Michaelis–Menten curve fitted on 8 models measured on both: y = 0.84681·(x − 0.0000) / (0.0000 + 1.8176 − x), R² = 0.48, cross-validated error 3.2 pp. It is used for 4 estimates.

| Estimated model | MMLU-ProX | SWE Multilingual | Source |
|---|---|---|---|
| [Qwen3.5-122B-A10B](https://benchgap.net/model/qwen3-5-122b-a10b.md) | 82.2% | 69.9% | estimated ± 3.2 pp, low confidence |
| [Qwen3.5-27B](https://benchgap.net/model/qwen3-5-27b.md) | 82.2% | 69.9% | estimated ± 3.2 pp, low confidence |
| [Qwen3.5-35B-A3B](https://benchgap.net/model/qwen3-5-35b-a3b.md) | 81.0% | 68.1% | estimated ± 3.2 pp, low confidence |
| [Qwen3 235B 2507](https://benchgap.net/model/qwen3-235b-2507.md) | 79.4% | 65.7% | estimated ± 3.2 pp, low confidence |

## Fitted on

| Model | MMLU-ProX | SWE Multilingual |
|---|---|---|
| [Claude Opus 4.5](https://benchgap.net/model/claude-opus-4-5.md) | 85.7% | 77.5% |
| [GLM-5](https://benchgap.net/model/glm-5.md) | 83.1% | 73.3% |
| [Kimi K2.5](https://benchgap.net/model/kimi-k2-5.md) | 82.3% | 73.0% |
| [Nemotron 3 Ultra](https://benchgap.net/model/nemotron-3-ultra.md) | 83.0% | 67.7% |
| [Qwen3.5 397B](https://benchgap.net/model/qwen3-5-397b.md) | 84.7% | 69.3% |
| [Qwen3.6 Plus](https://benchgap.net/model/qwen3-6-plus.md) | 84.7% | 73.8% |
| [Qwen3.7 Max](https://benchgap.net/model/qwen3-7-max.md) | 87.0% | 78.3% |
| [Qwen3.7 Plus](https://benchgap.net/model/qwen3-7-plus.md) | 85.4% | 75.8% |
