# Clinical Task Execution on MedAgentBench OOD v2

Source: [Clinical Task Execution on MedAgentBench OOD v2](<https://www.sota2.com/research/sota/clinical-task-execution-on-medagentbench-ood-v2>)

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

Data updated: 2026-05-29T09:53:30.940Z

Evaluation of model performance on Clinical Task Execution using the Out-of-Distribution version 2 split of the MedAgentBench dataset, measured by Accuracy.

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 | Clinical Task Execution |
| Dataset (including any stated split) | MedAgentBench OOD v2 |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 35 of 35 |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/clinical-task-execution-on-medagentbench-ood-v2.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| SkillX | 87.1 | Not recorded | 6a191dcfdbd7150506f82a5c |
| Evo-MedAgent | 86.7 | Not recorded | 6a191dcfdbd7150506f82a5c |
| SkillX | 86.7 | Not recorded | 6a191dcfdbd7150506f82a5c |
| ExpeL | 86.3 | Not recorded | 6a191dcfdbd7150506f82a5c |
| ExpeL | 86.1 | Not recorded | 6a191dcfdbd7150506f82a5c |
| GRASP | 86 | Not recorded | 6a191dcfdbd7150506f82a5c |
| ExpeL | 85 | Not recorded | 6a191dcfdbd7150506f82a5c |
| No skills | 84.9 | Not recorded | 6a191dcfdbd7150506f82a5c |
| ExpeL | 84.4 | Not recorded | 6a191dcfdbd7150506f82a5c |
| Batch Memory | 83.9 | Not recorded | 6a191dcfdbd7150506f82a5c |
| No skills | 83.6 | Not recorded | 6a191dcfdbd7150506f82a5c |
| Seq. Memory | 83.3 | Not recorded | 6a191dcfdbd7150506f82a5c |
| Evo-MedAgent | 82.8 | Not recorded | 6a191dcfdbd7150506f82a5c |
| GRASP | 82.8 | Not recorded | 6a191dcfdbd7150506f82a5c |
| No skills | 82.7 | Not recorded | 6a191dcfdbd7150506f82a5c |
| Seq. Memory | 82.3 | Not recorded | 6a191dcfdbd7150506f82a5c |
| Seq. Memory | 81.7 | Not recorded | 6a191dcfdbd7150506f82a5c |
| SkillX | 80.6 | Not recorded | 6a191dcfdbd7150506f82a5c |
| GRASP | 80 | Not recorded | 6a191dcfdbd7150506f82a5c |
| Evo-MedAgent | 80 | Not recorded | 6a191dcfdbd7150506f82a5c |
| GRASP | 80 | Not recorded | 6a191dcfdbd7150506f82a5c |
| Batch Memory | 78.8 | Not recorded | 6a191dcfdbd7150506f82a5c |
| Batch Memory | 78.3 | Not recorded | 6a191dcfdbd7150506f82a5c |
| ExpeL | 77.2 | Not recorded | 6a191dcfdbd7150506f82a5c |
| No skills | 76.3 | Not recorded | 6a191dcfdbd7150506f82a5c |
| No skills | 75.2 | Not recorded | 6a191dcfdbd7150506f82a5c |
| Batch Memory | 73.9 | Not recorded | 6a191dcfdbd7150506f82a5c |
| Evo-MedAgent | 72.2 | Not recorded | 6a191dcfdbd7150506f82a5c |
| GRASP | 67 | Not recorded | 6a191dcfdbd7150506f82a5c |
| SkillX | 66.1 | Not recorded | 6a191dcfdbd7150506f82a5c |
| Seq. Memory | 65.3 | Not recorded | 6a191dcfdbd7150506f82a5c |
| Batch Memory | 57.3 | Not recorded | 6a191dcfdbd7150506f82a5c |
| Seq. Memory | 56.1 | Not recorded | 6a191dcfdbd7150506f82a5c |
| Evo-MedAgent | 42.7 | Not recorded | 6a191dcfdbd7150506f82a5c |
| SkillX | 28 | Not recorded | 6a191dcfdbd7150506f82a5c |

### SkillX

Accuracy: 87.1. Result ID: 6a1961e3fbc373592c142f74.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69d48fc27d0172140ec745ee.md>)

[Implementation](<https://github.com/zjunlp/SkillX>)

Recorded evaluation configuration: {"Model":"gpt-oss-120b"}. This record does not assert independent reproduction.

### Evo-MedAgent

Accuracy: 86.7. Result ID: 6a1961e3fbc373592c142f73.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69e1be6e7d0172140ec7d61f.md>)

Recorded evaluation configuration: {"Model":"gpt-oss-120b"}. This record does not assert independent reproduction.

### SkillX

Accuracy: 86.7. Result ID: 6a1961e3fbc373592c142f90.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69d48fc27d0172140ec745ee.md>)

[Implementation](<https://github.com/zjunlp/SkillX>)

Recorded evaluation configuration: {"Model":"GPT-5.4 (low)"}. This record does not assert independent reproduction.

### ExpeL

Accuracy: 86.3. Result ID: 6a1961e3fbc373592c142f72.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa764a1e209131cea6c2eb.md>)

Recorded evaluation configuration: {"Model":"gpt-oss-120b"}. This record does not assert independent reproduction.

### ExpeL

Accuracy: 86.1. Result ID: 6a1961e3fbc373592c142f8e.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa764a1e209131cea6c2eb.md>)

Recorded evaluation configuration: {"Model":"GPT-5.4 (low)"}. This record does not assert independent reproduction.

### GRASP

Accuracy: 86. Result ID: 6a1961e3fbc373592c142f75.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"gpt-oss-120b"}. This record does not assert independent reproduction.

### ExpeL

Accuracy: 85. Result ID: 6a1961e3fbc373592c142f79.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa764a1e209131cea6c2eb.md>)

Recorded evaluation configuration: {"Model":"DeepSeek V4 Flash"}. This record does not assert independent reproduction.

### No skills

Accuracy: 84.9. Result ID: 6a1961e3fbc373592c142f84.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"GPT-4.1"}. This record does not assert independent reproduction.

### ExpeL

Accuracy: 84.4. Result ID: 6a1961e3fbc373592c142f87.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa764a1e209131cea6c2eb.md>)

Recorded evaluation configuration: {"Model":"GPT-4.1"}. This record does not assert independent reproduction.

### Batch Memory

Accuracy: 83.9. Result ID: 6a1961e3fbc373592c142f86.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"GPT-4.1"}. This record does not assert independent reproduction.

### No skills

Accuracy: 83.6. Result ID: 6a1961e3fbc373592c142f76.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"DeepSeek V4 Flash"}. This record does not assert independent reproduction.

### Seq. Memory

Accuracy: 83.3. Result ID: 6a1961e3fbc373592c142f85.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"GPT-4.1"}. This record does not assert independent reproduction.

### Evo-MedAgent

Accuracy: 82.8. Result ID: 6a1961e3fbc373592c142f88.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69e1be6e7d0172140ec7d61f.md>)

Recorded evaluation configuration: {"Model":"GPT-4.1"}. This record does not assert independent reproduction.

### GRASP

Accuracy: 82.8. Result ID: 6a1961e3fbc373592c142f8a.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"GPT-4.1"}. This record does not assert independent reproduction.

### No skills

Accuracy: 82.7. Result ID: 6a1961e3fbc373592c142f8b.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"GPT-5.4 (low)"}. This record does not assert independent reproduction.

### Seq. Memory

Accuracy: 82.3. Result ID: 6a1961e3fbc373592c142f70.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"gpt-oss-120b"}. This record does not assert independent reproduction.

### Seq. Memory

Accuracy: 81.7. Result ID: 6a1961e3fbc373592c142f8c.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"GPT-5.4 (low)"}. This record does not assert independent reproduction.

### SkillX

Accuracy: 80.6. Result ID: 6a1961e3fbc373592c142f89.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69d48fc27d0172140ec745ee.md>)

[Implementation](<https://github.com/zjunlp/SkillX>)

Recorded evaluation configuration: {"Model":"GPT-4.1"}. This record does not assert independent reproduction.

### GRASP

Accuracy: 80. Result ID: 6a1961e3fbc373592c142f83.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"Gemini 3.1 Flash Lite"}. This record does not assert independent reproduction.

### Evo-MedAgent

Accuracy: 80. Result ID: 6a1961e3fbc373592c142f8f.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69e1be6e7d0172140ec7d61f.md>)

Recorded evaluation configuration: {"Model":"GPT-5.4 (low)"}. This record does not assert independent reproduction.

### GRASP

Accuracy: 80. Result ID: 6a1961e3fbc373592c142f91.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"GPT-5.4 (low)"}. This record does not assert independent reproduction.

### Batch Memory

Accuracy: 78.8. Result ID: 6a1961e3fbc373592c142f71.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"gpt-oss-120b"}. This record does not assert independent reproduction.

### Batch Memory

Accuracy: 78.3. Result ID: 6a1961e3fbc373592c142f8d.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"GPT-5.4 (low)"}. This record does not assert independent reproduction.

### ExpeL

Accuracy: 77.2. Result ID: 6a1961e3fbc373592c142f80.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa764a1e209131cea6c2eb.md>)

Recorded evaluation configuration: {"Model":"Gemini 3.1 Flash Lite"}. This record does not assert independent reproduction.

### No skills

Accuracy: 76.3. Result ID: 6a1961e3fbc373592c142f7d.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"Gemini 3.1 Flash Lite"}. This record does not assert independent reproduction.

### No skills

Accuracy: 75.2. Result ID: 6a1961e3fbc373592c142f6f.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"gpt-oss-120b"}. This record does not assert independent reproduction.

### Batch Memory

Accuracy: 73.9. Result ID: 6a1961e3fbc373592c142f7f.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"Gemini 3.1 Flash Lite"}. This record does not assert independent reproduction.

### Evo-MedAgent

Accuracy: 72.2. Result ID: 6a1961e3fbc373592c142f81.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69e1be6e7d0172140ec7d61f.md>)

Recorded evaluation configuration: {"Model":"Gemini 3.1 Flash Lite"}. This record does not assert independent reproduction.

### GRASP

Accuracy: 67. Result ID: 6a1961e3fbc373592c142f7c.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"DeepSeek V4 Flash"}. This record does not assert independent reproduction.

### SkillX

Accuracy: 66.1. Result ID: 6a1961e3fbc373592c142f82.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69d48fc27d0172140ec745ee.md>)

[Implementation](<https://github.com/zjunlp/SkillX>)

Recorded evaluation configuration: {"Model":"Gemini 3.1 Flash Lite"}. This record does not assert independent reproduction.

### Seq. Memory

Accuracy: 65.3. Result ID: 6a1961e3fbc373592c142f77.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"DeepSeek V4 Flash"}. This record does not assert independent reproduction.

### Batch Memory

Accuracy: 57.3. Result ID: 6a1961e3fbc373592c142f78.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"DeepSeek V4 Flash"}. This record does not assert independent reproduction.

### Seq. Memory

Accuracy: 56.1. Result ID: 6a1961e3fbc373592c142f7e.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

Recorded evaluation configuration: {"Model":"Gemini 3.1 Flash Lite"}. This record does not assert independent reproduction.

### Evo-MedAgent

Accuracy: 42.7. Result ID: 6a1961e3fbc373592c142f7a.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69e1be6e7d0172140ec7d61f.md>)

Recorded evaluation configuration: {"Model":"DeepSeek V4 Flash"}. This record does not assert independent reproduction.

### SkillX

Accuracy: 28. Result ID: 6a1961e3fbc373592c142f7b.

[Reporting paper](<https://www.sota2.com/research/paper/6a191dcfdbd7150506f82a5c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69d48fc27d0172140ec745ee.md>)

[Implementation](<https://github.com/zjunlp/SkillX>)

Recorded evaluation configuration: {"Model":"DeepSeek V4 Flash"}. This record does not assert independent reproduction.

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