# DeepSWE → SciCode

Calibration · https://benchgap.net/calibration/deepswe/current/to/scicode/current · data retrieved 2026-11-18

SciCode is estimated from DeepSWE with a Michaelis–Menten + offset curve fitted on 5 models measured on both: y = 0.1843 + 2.0000·x / (2.70295 + x), R² = 0.64, cross-validated error 7.6 pp. It is used for 14 estimates.

| Estimated model | DeepSWE | SciCode | Source |
|---|---|---|---|
| [Qwen4 Max](https://benchgap.net/model/qwen4-max.md) | 75.7% | 62.2% | estimated ± 7.6 pp, low confidence |
| [Hy4](https://benchgap.net/model/hy4.md) | 74.9% | 61.8% | estimated ± 7.6 pp, low confidence |
| [Pareto 26.9](https://benchgap.net/model/pareto-26-9.md) | 74.0% | 61.4% | estimated ± 7.6 pp, low confidence |
| [MAI Code 2](https://benchgap.net/model/mai-code-2.md) | 73.2% | 61.0% | estimated ± 7.6 pp, low confidence |
| [SWE-2](https://benchgap.net/model/swe-2.md) | 73.0% | 61.0% | estimated ± 7.6 pp, low confidence |
| [Particle-1](https://benchgap.net/model/particle-1.md) | 72.6% | 60.8% | estimated ± 7.6 pp, low confidence |
| [ERNIE 5.5 Flash Preview](https://benchgap.net/model/ernie-5-5-flash-preview.md) | 71.2% | 60.1% | estimated ± 7.6 pp, low confidence |
| [Pareto 26.10 Preview](https://benchgap.net/model/pareto-26-10-preview.md) | 69.9% | 59.5% | estimated ± 7.6 pp, low confidence |
| [Hy4 flash](https://benchgap.net/model/hy4-flash.md) | 64.6% | 57.0% | estimated ± 7.6 pp, low confidence |
| [LongCat-2.5](https://benchgap.net/model/longcat-2-5.md) | 64.6% | 57.0% | estimated ± 7.6 pp, low confidence |
| [Ultramarine 1 Flash](https://benchgap.net/model/ultramarine-1-flash.md) | 61.5% | 55.5% | estimated ± 7.6 pp, low confidence |
| [SWE-2 Flash](https://benchgap.net/model/swe-2-flash.md) | 60.1% | 54.8% | estimated ± 7.6 pp, low confidence |
| [MAI Code 2 Flash](https://benchgap.net/model/mai-code-2-flash.md) | 59.9% | 54.7% | estimated ± 7.6 pp, low confidence |
| [Mercury 3](https://benchgap.net/model/mercury-3.md) | 46.3% | 47.7% | estimated ± 7.6 pp, low confidence |

## Fitted on

| Model | DeepSWE | SciCode |
|---|---|---|
| [Beam 1.0 501B-A23B](https://benchgap.net/model/reflection-beam.md) | 43.3% | 49.2% |
| [Beam 1.0 501B-A23B FP8](https://benchgap.net/model/reflection-beam-fp8.md) | 43.0% | 48.3% |
| [Beam 1.0 501B-A23B NVFP4](https://benchgap.net/model/reflection-beam-nvfp4.md) | 42.7% | 46.9% |
| [Ripple-65B Preview](https://benchgap.net/model/ripple-65b-preview.md) | 44.3% | 39.4% |
| [Step 5 Preview](https://benchgap.net/model/step-5-preview.md) | 67.7% | 58.9% |
