# Scientific Reasoning on GPQA Diamond (Accuracy MV, Accuracy WMV, Required Tokens)

Source: [Scientific Reasoning on GPQA Diamond (Accuracy MV, Accuracy WMV, Required Tokens)](<https://www.sota2.com/research/sota/scientific-reasoning-on-gpqa-diamond-accuracy-mv-accuracy-wmv-required-tokens>)

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

Data updated: 2026-07-01T08:01:56.456Z

Evaluation of model performance on the GPQA Diamond dataset for Scientific Reasoning, reporting results for Most Valuable Answer (MV) accuracy, Worst-case Most Valuable Answer (WMV) accuracy, and the average number of required tokens.

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 | Scientific Reasoning |
| Dataset (including any stated split) | GPQA Diamond |
| Metric | Accuracy (MV) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 9 of 9 |
| 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 (MV)](<https://www.sota2.com/research/sota/scientific-reasoning-on-gpqa-diamond-accuracy-mv-accuracy-wmv-required-tokens.md?metric=Accuracy%20(MV)>): Higher is better; unit: PERCENT.

- [Accuracy (WMV)](<https://www.sota2.com/research/sota/scientific-reasoning-on-gpqa-diamond-accuracy-mv-accuracy-wmv-required-tokens.md?metric=Accuracy%20(WMV)>): Higher is better; unit: PERCENT.

- [Required Tokens](<https://www.sota2.com/research/sota/scientific-reasoning-on-gpqa-diamond-accuracy-mv-accuracy-wmv-required-tokens.md?metric=Required%20Tokens>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Accuracy (MV) | Result date | Report paper |
| --- | --- | --- | --- |
| DeepConf | 72.7 | Not recorded | 6a449f12e0224bd4f49930df |
| Flex-Fork-think | 70.7 | Not recorded | 6a449f12e0224bd4f49930df |
| DeepConf | 67.3 | Not recorded | 6a449f12e0224bd4f49930df |
| ASC | 67.1 | Not recorded | 6a449f12e0224bd4f49930df |
| Flex-Fork-think | 63.1 | Not recorded | 6a449f12e0224bd4f49930df |
| DeepConf | 62.8 | Not recorded | 6a449f12e0224bd4f49930df |
| Flex-Fork-think | 62.1 | Not recorded | 6a449f12e0224bd4f49930df |
| ASC | 61.7 | Not recorded | 6a449f12e0224bd4f49930df |
| ASC | 58.1 | Not recorded | 6a449f12e0224bd4f49930df |

### DeepConf

Accuracy (MV): 72.7. Result ID: 6a44c97373d097d845bef7b1.

[Reporting paper](<https://www.sota2.com/research/paper/6a449f12e0224bd4f49930df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7ad41e209131cea855e8.md>)

Recorded evaluation configuration: {"Backbone":"Phi-4-RP-14B","branching factor n":"32"}. This record does not assert independent reproduction.

### Flex-Fork-think

Accuracy (MV): 70.7. Result ID: 6a44c97373d097d845bef7b2.

[Reporting paper](<https://www.sota2.com/research/paper/6a449f12e0224bd4f49930df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a449f12e0224bd4f49930df.md>)

Recorded evaluation configuration: {"Backbone":"Phi-4-RP-14B","branching factor n":"32","seed path length l0":"2,048"}. This record does not assert independent reproduction.

### DeepConf

Accuracy (MV): 67.3. Result ID: 6a44c97373d097d845bef7ae.

[Reporting paper](<https://www.sota2.com/research/paper/6a449f12e0224bd4f49930df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7ad41e209131cea855e8.md>)

Recorded evaluation configuration: {"Backbone":"DeepSeek-8B","branching factor n":"32"}. This record does not assert independent reproduction.

### ASC

Accuracy (MV): 67.1. Result ID: 6a44c97373d097d845bef7ad.

[Reporting paper](<https://www.sota2.com/research/paper/6a449f12e0224bd4f49930df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa713f1e209131cea5017a.md>)

Recorded evaluation configuration: {"Backbone":"DeepSeek-8B","branching factor n":"32"}. This record does not assert independent reproduction.

### Flex-Fork-think

Accuracy (MV): 63.1. Result ID: 6a44c97373d097d845bef7ac.

[Reporting paper](<https://www.sota2.com/research/paper/6a449f12e0224bd4f49930df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a449f12e0224bd4f49930df.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-8B","branching factor n":"32","seed path length l0":"2,048"}. This record does not assert independent reproduction.

### DeepConf

Accuracy (MV): 62.8. Result ID: 6a44c97373d097d845bef7ab.

[Reporting paper](<https://www.sota2.com/research/paper/6a449f12e0224bd4f49930df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7ad41e209131cea855e8.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-8B","branching factor n":"32"}. This record does not assert independent reproduction.

### Flex-Fork-think

Accuracy (MV): 62.1. Result ID: 6a44c97373d097d845bef7af.

[Reporting paper](<https://www.sota2.com/research/paper/6a449f12e0224bd4f49930df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a449f12e0224bd4f49930df.md>)

Recorded evaluation configuration: {"Backbone":"DeepSeek-8B","branching factor n":"32","seed path length l0":"2,048"}. This record does not assert independent reproduction.

### ASC

Accuracy (MV): 61.7. Result ID: 6a44c97373d097d845bef7aa.

[Reporting paper](<https://www.sota2.com/research/paper/6a449f12e0224bd4f49930df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa713f1e209131cea5017a.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-8B","branching factor n":"32"}. This record does not assert independent reproduction.

### ASC

Accuracy (MV): 58.1. Result ID: 6a44c97373d097d845bef7b0.

[Reporting paper](<https://www.sota2.com/research/paper/6a449f12e0224bd4f49930df.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa713f1e209131cea5017a.md>)

Recorded evaluation configuration: {"Backbone":"Phi-4-RP-14B","branching factor n":"32"}. This record does not assert independent reproduction.

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