# Medical Reasoning on MMLU (val)

Source: [Medical Reasoning on MMLU (val)](<https://www.sota2.com/research/sota/medical-reasoning-on-mmlu-val>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-06-23T11:35:20.450Z

Evaluation of model performance on the Medical Reasoning task using the validation split of the standard MMLU benchmark suite, reporting accuracy across specific medical domains.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Medical Reasoning |
| Dataset (including any stated split) | MMLU (val) |
| Metric | Anatomy (AN) Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 10 of 10 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Anatomy (AN) Accuracy](<https://www.sota2.com/research/sota/medical-reasoning-on-mmlu-val.md?metric=Anatomy%20(AN)%20Accuracy>): Higher is better; unit: PERCENT.

- [Clinical Knowledge (CK) Accuracy](<https://www.sota2.com/research/sota/medical-reasoning-on-mmlu-val.md?metric=Clinical%20Knowledge%20(CK)%20Accuracy>): Higher is better; unit: PERCENT.

- [College Biology (CB) Accuracy](<https://www.sota2.com/research/sota/medical-reasoning-on-mmlu-val.md?metric=College%20Biology%20(CB)%20Accuracy>): Higher is better; unit: PERCENT.

- [College Medicine (CM) Accuracy](<https://www.sota2.com/research/sota/medical-reasoning-on-mmlu-val.md?metric=College%20Medicine%20(CM)%20Accuracy>): Higher is better; unit: PERCENT.

- [Medical Genetics (MG) Accuracy](<https://www.sota2.com/research/sota/medical-reasoning-on-mmlu-val.md?metric=Medical%20Genetics%20(MG)%20Accuracy>): Higher is better; unit: PERCENT.

- [Professional Medicine (PM) Accuracy](<https://www.sota2.com/research/sota/medical-reasoning-on-mmlu-val.md?metric=Professional%20Medicine%20(PM)%20Accuracy>): Higher is better; unit: PERCENT.

- [Average Accuracy (MMLU Medical Reasoning)](<https://www.sota2.com/research/sota/medical-reasoning-on-mmlu-val.md?metric=Average%20Accuracy%20(MMLU%20Medical%20Reasoning)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Anatomy (AN) Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Baichuan-M1-14B | 77.3 | Not recorded | 68fa7bc71e209131cea8a6a6 |
| Qwen3-8B+E2C-(SFT+RL) | 77 | Not recorded | 68fa7bc71e209131cea8a6a6 |
| ReasonMed-7B | 75.6 | Not recorded | 68fa7bc71e209131cea8a6a6 |
| Qwen3-8B + GRPO | 75 | Not recorded | 68fa7bc71e209131cea8a6a6 |
| Qwen3-8B + E2C-SFT + EF-SFT | 72.3 | Not recorded | 68fa7bc71e209131cea8a6a6 |
| HuatuoGPT-o1-7B | 71.9 | Not recorded | 68fa7bc71e209131cea8a6a6 |
| Qwen3-8B + standard SFT | 68.8 | Not recorded | 68fa7bc71e209131cea8a6a6 |
| Qwen3-8B | 68 | Not recorded | 68fa7bc71e209131cea8a6a6 |
| Llama3.1-8B + E2C-SFT + EF-SFT | 61.8 | Not recorded | 68fa7bc71e209131cea8a6a6 |
| Llama3.1-8B + ReasonMed SFT | 45.9 | Not recorded | 68fa7bc71e209131cea8a6a6 |

### Baichuan-M1-14B

Anatomy (AN) Accuracy: 77.3. Result ID: 6a3a6f60fbc373592c37f1d6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bc71e209131cea8a6a6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77481e209131cea723dd.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Qwen3-8B+E2C-(SFT+RL)

Anatomy (AN) Accuracy: 77. Result ID: 6a3a6f60fbc373592c37f1da.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bc71e209131cea8a6a6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bc71e209131cea8a6a6.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### ReasonMed-7B

Anatomy (AN) Accuracy: 75.6. Result ID: 6a3a6f60fbc373592c37f1d7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bc71e209131cea8a6a6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c021e209131cea8bd65.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Qwen3-8B + GRPO

Anatomy (AN) Accuracy: 75. Result ID: 6a3a6f60fbc373592c37f1d9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bc71e209131cea8a6a6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75001e209131cea652ab.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Qwen3-8B + E2C-SFT + EF-SFT

Anatomy (AN) Accuracy: 72.3. Result ID: 6a3a6f60fbc373592c37f1dc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bc71e209131cea8a6a6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bc71e209131cea8a6a6.md>)

Recorded evaluation configuration: {"#Med-Tokens":"10M"}. This record does not assert independent reproduction.

### HuatuoGPT-o1-7B

Anatomy (AN) Accuracy: 71.9. Result ID: 6a3a6f60fbc373592c37f1d5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bc71e209131cea8a6a6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73691e209131cea5c514.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Qwen3-8B + standard SFT

Anatomy (AN) Accuracy: 68.8. Result ID: 6a3a6f60fbc373592c37f1db.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bc71e209131cea8a6a6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa788e1e209131cea79116.md>)

Recorded evaluation configuration: {"#Med-Tokens":"286M"}. This record does not assert independent reproduction.

### Qwen3-8B

Anatomy (AN) Accuracy: 68. Result ID: 6a3a6f60fbc373592c37f1d8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bc71e209131cea8a6a6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa788e1e209131cea79116.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Llama3.1-8B + E2C-SFT + EF-SFT

Anatomy (AN) Accuracy: 61.8. Result ID: 6a3a6f60fbc373592c37f1de.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bc71e209131cea8a6a6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bc71e209131cea8a6a6.md>)

Recorded evaluation configuration: {"#Med-Tokens":"10M"}. This record does not assert independent reproduction.

### Llama3.1-8B + ReasonMed SFT

Anatomy (AN) Accuracy: 45.9. Result ID: 6a3a6f60fbc373592c37f1dd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bc71e209131cea8a6a6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75d11e209131cea6a028.md>)

[Implementation](<https://github.com/meta-llama/llama-agentic-system>)

Recorded evaluation configuration: {"#Med-Tokens":"286M"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.
