# Report Generation on DeepResearch Bench 2025 (test)

Source: [Report Generation on DeepResearch Bench 2025 (test)](<https://www.sota2.com/research/sota/report-generation-on-deepresearch-bench-2025-test>)

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

Data updated: 2026-02-26T05:39:40.837Z

Evaluation of model performance on generating research reports using the test split of the DeepResearch Bench 2025 dataset, focusing on metrics like Comprehensiveness, Depth, Instruction Following, Readability, and Overall score.

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 | Report Generation |
| Dataset (including any stated split) | DeepResearch Bench 2025 (test) |
| Metric | Comprehensiveness |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 16 of 16 |
| 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

- [Comprehensiveness](<https://www.sota2.com/research/sota/report-generation-on-deepresearch-bench-2025-test.md?metric=Comprehensiveness>): Higher is better; unit: PERCENT.

- [Depth](<https://www.sota2.com/research/sota/report-generation-on-deepresearch-bench-2025-test.md?metric=Depth>): Higher is better; unit: PERCENT.

- [Instruction Following](<https://www.sota2.com/research/sota/report-generation-on-deepresearch-bench-2025-test.md?metric=Instruction%20Following>): Higher is better; unit: PERCENT.

- [Readability](<https://www.sota2.com/research/sota/report-generation-on-deepresearch-bench-2025-test.md?metric=Readability>): Higher is better; unit: PERCENT.

- [Overall Score](<https://www.sota2.com/research/sota/report-generation-on-deepresearch-bench-2025-test.md?metric=Overall%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Comprehensiveness | Result date | Report paper |
| --- | --- | --- | --- |
| Gemini DeepResearch | 49.5 | Not recorded | 6982d2eea8b01e76cc8e990b |
| Tongyi-DeepResearch | 48.3 | Not recorded | 6982d2eea8b01e76cc8e990b |
| OpenAI DeepResearch | 46.5 | Not recorded | 6982d2eea8b01e76cc8e990b |
| Claude Research | 45.3 | Not recorded | 6982d2eea8b01e76cc8e990b |
| WebWeaver-Qwen3-30B-A3B | 45.2 | Not recorded | 6982d2eea8b01e76cc8e990b |
| Tongyi-DeepResearch | 43.4 | Not recorded | 6982d2eea8b01e76cc8e990b |
| DRTulu-Qwen3-8B-RL | 41.7 | Not recorded | 6982d2eea8b01e76cc8e990b |
| Qwen3-30B-A3B | 41.5 | Not recorded | 6982d2eea8b01e76cc8e990b |
| Tongyi-DeepResearch | 41.1 | Not recorded | 6982d2eea8b01e76cc8e990b |
| Tongyi-DeepResearch | 40.5 | Not recorded | 6982d2eea8b01e76cc8e990b |
| Tongyi-DeepResearch | 40.5 | Not recorded | 6982d2eea8b01e76cc8e990b |
| Tongyi-DeepResearch | 39.5 | Not recorded | 6982d2eea8b01e76cc8e990b |
| WebThinker-Qwen2.5-32B-DPO | 39.4 | Not recorded | 6982d2eea8b01e76cc8e990b |
| Tongyi-DeepResearch | 38.9 | Not recorded | 6982d2eea8b01e76cc8e990b |
| Qwen3-30B-A3B | 38.5 | Not recorded | 6982d2eea8b01e76cc8e990b |
| Qwen3-30B-A3B | 33.8 | Not recorded | 6982d2eea8b01e76cc8e990b |

### Gemini DeepResearch

Comprehensiveness: 49.5. Result ID: 69964f165698665ec493c2e3.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

Recorded evaluation configuration: {"Workflow":"N/A","Rubric Strategy":"N/A"}. This record does not assert independent reproduction.

### Tongyi-DeepResearch

Comprehensiveness: 48.3. Result ID: 69964f165698665ec493c2f0.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

Recorded evaluation configuration: {"Workflow":"MaMs","Rubric Strategy":"Rubric Generator Trained by RL"}. This record does not assert independent reproduction.

### OpenAI DeepResearch

Comprehensiveness: 46.5. Result ID: 69964f165698665ec493c2e1.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

Recorded evaluation configuration: {"Workflow":"N/A","Rubric Strategy":"N/A"}. This record does not assert independent reproduction.

### Claude Research

Comprehensiveness: 45.3. Result ID: 69964f165698665ec493c2e2.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

Recorded evaluation configuration: {"Workflow":"N/A","Rubric Strategy":"N/A"}. This record does not assert independent reproduction.

### WebWeaver-Qwen3-30B-A3B

Comprehensiveness: 45.2. Result ID: 69964f165698665ec493c2e6.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bf41e209131cea8b8da.md>)

Recorded evaluation configuration: {"Workflow":"WebWeaver","Rubric Strategy":"N/A"}. This record does not assert independent reproduction.

### Tongyi-DeepResearch

Comprehensiveness: 43.4. Result ID: 69964f165698665ec493c2eb.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

Recorded evaluation configuration: {"Workflow":"ReAct","Rubric Strategy":"Rubric Generator Trained by RL"}. This record does not assert independent reproduction.

### DRTulu-Qwen3-8B-RL

Comprehensiveness: 41.7. Result ID: 69964f165698665ec493c2e5.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69253885d6b0dd334292f936.md>)

[Implementation](<https://github.com/rlresearch/dr-tulu>)

Recorded evaluation configuration: {"Workflow":"ReAct","Rubric Strategy":"Self-Evolving Rubrics"}. This record does not assert independent reproduction.

### Qwen3-30B-A3B

Comprehensiveness: 41.5. Result ID: 69964f165698665ec493c2e9.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

Recorded evaluation configuration: {"Workflow":"MaMs","Rubric Strategy":"Rubric Generator Trained by RL"}. This record does not assert independent reproduction.

### Tongyi-DeepResearch

Comprehensiveness: 41.1. Result ID: 69964f165698665ec493c2ee.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

Recorded evaluation configuration: {"Workflow":"MaMs","Rubric Strategy":"GPT-5 Generated Rubrics"}. This record does not assert independent reproduction.

### Tongyi-DeepResearch

Comprehensiveness: 40.5. Result ID: 69964f165698665ec493c2ed.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

Recorded evaluation configuration: {"Workflow":"MaMs","Rubric Strategy":"Human-defined General Rubrics"}. This record does not assert independent reproduction.

### Tongyi-DeepResearch

Comprehensiveness: 40.5. Result ID: 69964f165698665ec493c2ef.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

Recorded evaluation configuration: {"Workflow":"MaMs","Rubric Strategy":"Rubric Generator Trained by SFT"}. This record does not assert independent reproduction.

### Tongyi-DeepResearch

Comprehensiveness: 39.5. Result ID: 69964f165698665ec493c2ea.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

Recorded evaluation configuration: {"Workflow":"ReAct","Rubric Strategy":"N/A"}. This record does not assert independent reproduction.

### WebThinker-Qwen2.5-32B-DPO

Comprehensiveness: 39.4. Result ID: 69964f165698665ec493c2e4.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c191e209131cea8c4d7.md>)

Recorded evaluation configuration: {"Workflow":"WebThinker","Rubric Strategy":"N/A"}. This record does not assert independent reproduction.

### Tongyi-DeepResearch

Comprehensiveness: 38.9. Result ID: 69964f165698665ec493c2ec.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

Recorded evaluation configuration: {"Workflow":"MaMs","Rubric Strategy":"N/A"}. This record does not assert independent reproduction.

### Qwen3-30B-A3B

Comprehensiveness: 38.5. Result ID: 69964f165698665ec493c2e8.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

Recorded evaluation configuration: {"Workflow":"ReAct","Rubric Strategy":"Rubric Generator Trained by RL"}. This record does not assert independent reproduction.

### Qwen3-30B-A3B

Comprehensiveness: 33.8. Result ID: 69964f165698665ec493c2e7.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6982d2eea8b01e76cc8e990b.md>)

Recorded evaluation configuration: {"Workflow":"ReAct","Rubric Strategy":"N/A"}. This record does not assert independent reproduction.

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