# General AI Assistant Reasoning on GAIA (File/Reasoning/Others)

Source: [General AI Assistant Reasoning on GAIA (File/Reasoning/Others)](<https://www.sota2.com/research/sota/general-ai-assistant-reasoning-on-gaia-file-reasoning-others>)

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

Data updated: 2026-02-26T05:30:46.876Z

Evaluation of General AI Assistant Reasoning capabilities specifically on the File/Reasoning/Others subset of the GAIA dataset, measured using Accuracy and Gain metrics.

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 | General AI Assistant Reasoning |
| Dataset (including any stated split) | GAIA (File/Reasoning/Others) |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 18 of 18 |
| 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/general-ai-assistant-reasoning-on-gaia-file-reasoning-others.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Final Gain](<https://www.sota2.com/research/sota/general-ai-assistant-reasoning-on-gaia-file-reasoning-others.md?metric=Final%20Gain>): Higher is better; unit: SCALAR.

- [Best Gain](<https://www.sota2.com/research/sota/general-ai-assistant-reasoning-on-gaia-file-reasoning-others.md?metric=Best%20Gain>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Claude-3.7 | 56.21 | Not recorded | 69730077b2be0ed85a78c980 |
| Claude-3.7 | 54.92 | Not recorded | 69730077b2be0ed85a78c980 |
| Claude-3.7 | 54.92 | Not recorded | 69730077b2be0ed85a78c980 |
| Claude-3.7 | 54.92 | Not recorded | 69730077b2be0ed85a78c980 |
| Claude-3.7 | 53.57 | Not recorded | 69730077b2be0ed85a78c980 |
| Claude-3.7 | 53.57 | Not recorded | 69730077b2be0ed85a78c980 |
| GPT-4.1 | 33.33 | Not recorded | 69730077b2be0ed85a78c980 |
| GPT-4.1 | 33.33 | Not recorded | 69730077b2be0ed85a78c980 |
| GPT-4.1 | 33.33 | Not recorded | 69730077b2be0ed85a78c980 |
| GPT-4.1 | 33.33 | Not recorded | 69730077b2be0ed85a78c980 |
| GPT-4.1 | 33.33 | Not recorded | 69730077b2be0ed85a78c980 |
| DeepVerifier-8B | 30.85 | Not recorded | 69730077b2be0ed85a78c980 |
| DeepVerifier-8B | 30.85 | Not recorded | 69730077b2be0ed85a78c980 |
| DeepVerifier-8B | 30.85 | Not recorded | 69730077b2be0ed85a78c980 |
| DeepVerifier-8B | 30.85 | Not recorded | 69730077b2be0ed85a78c980 |
| DeepVerifier-8B | 30.85 | Not recorded | 69730077b2be0ed85a78c980 |
| GPT-4.1 | 30.67 | Not recorded | 69730077b2be0ed85a78c980 |
| DeepVerifier-8B | 26.81 | Not recorded | 69730077b2be0ed85a78c980 |

### Claude-3.7

Accuracy: 56.21. Result ID: 699624025698665ec464a2ee.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"Claude-3.7-Sonnet","Feedback Rounds":4}. This record does not assert independent reproduction.

### Claude-3.7

Accuracy: 54.92. Result ID: 699624025698665ec464a2ef.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"Claude-3.7-Sonnet","Feedback Rounds":6}. This record does not assert independent reproduction.

### Claude-3.7

Accuracy: 54.92. Result ID: 699624025698665ec464a2f0.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"Claude-3.7-Sonnet","Feedback Rounds":8}. This record does not assert independent reproduction.

### Claude-3.7

Accuracy: 54.92. Result ID: 699624025698665ec464a2f1.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"Claude-3.7-Sonnet","Feedback Rounds":10}. This record does not assert independent reproduction.

### Claude-3.7

Accuracy: 53.57. Result ID: 699624025698665ec464a2ec.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"Claude-3.7-Sonnet","Feedback Rounds":0}. This record does not assert independent reproduction.

### Claude-3.7

Accuracy: 53.57. Result ID: 699624025698665ec464a2ed.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"Claude-3.7-Sonnet","Feedback Rounds":2}. This record does not assert independent reproduction.

### GPT-4.1

Accuracy: 33.33. Result ID: 699624025698665ec464a2f3.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"GPT-4.1","Feedback Rounds":2}. This record does not assert independent reproduction.

### GPT-4.1

Accuracy: 33.33. Result ID: 699624025698665ec464a2f4.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"GPT-4.1","Feedback Rounds":4}. This record does not assert independent reproduction.

### GPT-4.1

Accuracy: 33.33. Result ID: 699624025698665ec464a2f5.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"GPT-4.1","Feedback Rounds":6}. This record does not assert independent reproduction.

### GPT-4.1

Accuracy: 33.33. Result ID: 699624025698665ec464a2f6.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"GPT-4.1","Feedback Rounds":8}. This record does not assert independent reproduction.

### GPT-4.1

Accuracy: 33.33. Result ID: 699624025698665ec464a2f7.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"GPT-4.1","Feedback Rounds":10}. This record does not assert independent reproduction.

### DeepVerifier-8B

Accuracy: 30.85. Result ID: 699624025698665ec464a2f9.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"Llama-3-8B-Base","Feedback Rounds":2}. This record does not assert independent reproduction.

### DeepVerifier-8B

Accuracy: 30.85. Result ID: 699624025698665ec464a2fa.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"Llama-3-8B-Base","Feedback Rounds":4}. This record does not assert independent reproduction.

### DeepVerifier-8B

Accuracy: 30.85. Result ID: 699624025698665ec464a2fb.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"Llama-3-8B-Base","Feedback Rounds":6}. This record does not assert independent reproduction.

### DeepVerifier-8B

Accuracy: 30.85. Result ID: 699624025698665ec464a2fc.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"Llama-3-8B-Base","Feedback Rounds":8}. This record does not assert independent reproduction.

### DeepVerifier-8B

Accuracy: 30.85. Result ID: 699624025698665ec464a2fd.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"Llama-3-8B-Base","Feedback Rounds":10}. This record does not assert independent reproduction.

### GPT-4.1

Accuracy: 30.67. Result ID: 699624025698665ec464a2f2.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"GPT-4.1","Feedback Rounds":0}. This record does not assert independent reproduction.

### DeepVerifier-8B

Accuracy: 26.81. Result ID: 699624025698665ec464a2f8.

[Reporting paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69730077b2be0ed85a78c980.md>)

Recorded evaluation configuration: {"Backbone":"Llama-3-8B-Base","Feedback Rounds":0}. This record does not assert independent reproduction.

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