# Offline Action Consistency Diagnostic on AITW

Source: [Offline Action Consistency Diagnostic on AITW](<https://www.sota2.com/research/sota/offline-action-consistency-diagnostic-on-aitw>)

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

Data updated: 2026-06-24T08:34:46.383Z

Evaluation of model performance on the AITW dataset for Offline Action Consistency Diagnostic, measured by action 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 | Offline Action Consistency Diagnostic |
| Dataset (including any stated split) | AITW |
| Metric | Action Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 15 of 15 |
| 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

- [Action Accuracy](<https://www.sota2.com/research/sota/offline-action-consistency-diagnostic-on-aitw.md?metric=Action%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Action Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Qwen2.5-VL Agent + VisCritic | 74.2 | Not recorded | 6a3b64d8206542251d7c993a |
| Qwen2.5-VL Agent + GUI-Critic-R1 | 73.3 | Not recorded | 6a3b64d8206542251d7c993a |
| Qwen2.5-VL Agent + GuidNav | 73 | Not recorded | 6a3b64d8206542251d7c993a |
| Qwen2.5-VL Agent + GUI-PRA | 72.5 | Not recorded | 6a3b64d8206542251d7c993a |
| Qwen2.5-VL Agent | 70.4 | Not recorded | 6a3b64d8206542251d7c993a |
| ShowUI + VisCritic | 69.8 | Not recorded | 6a3b64d8206542251d7c993a |
| ShowUI + GUI-PRA | 68.8 | Not recorded | 6a3b64d8206542251d7c993a |
| CogAgent + VisCritic | 67.1 | Not recorded | 6a3b64d8206542251d7c993a |
| ShowUI | 66.7 | Not recorded | 6a3b64d8206542251d7c993a |
| SeeClick + GUI-Critic-R1 | 64.8 | Not recorded | 6a3b64d8206542251d7c993a |
| CogAgent | 64.8 | Not recorded | 6a3b64d8206542251d7c993a |
| SeeClick + GUI-PRA | 64.5 | Not recorded | 6a3b64d8206542251d7c993a |
| SeeClick + VisCritic | 64.5 | Not recorded | 6a3b64d8206542251d7c993a |
| SeeClick + BacktrackAgent | 64.2 | Not recorded | 6a3b64d8206542251d7c993a |
| SeeClick | 62.1 | Not recorded | 6a3b64d8206542251d7c993a |

### Qwen2.5-VL Agent + VisCritic

Action Accuracy: 74.2. Result ID: 6a3b969cfbc373592c44b232.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

Recorded evaluation configuration: {"base\_agent":"Qwen2.5-VL Agent","critic\_module":"VisCritic"}. This record does not assert independent reproduction.

### Qwen2.5-VL Agent + GUI-Critic-R1

Action Accuracy: 73.3. Result ID: 6a3b969cfbc373592c44b231.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

Recorded evaluation configuration: {"base\_agent":"Qwen2.5-VL Agent","critic\_module":"GUI-Critic-R1"}. This record does not assert independent reproduction.

### Qwen2.5-VL Agent + GuidNav

Action Accuracy: 73. Result ID: 6a3b969cfbc373592c44b230.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

Recorded evaluation configuration: {"base\_agent":"Qwen2.5-VL Agent","critic\_module":"GuidNav"}. This record does not assert independent reproduction.

### Qwen2.5-VL Agent + GUI-PRA

Action Accuracy: 72.5. Result ID: 6a3b969cfbc373592c44b22f.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

Recorded evaluation configuration: {"base\_agent":"Qwen2.5-VL Agent","critic\_module":"GUI-PRA"}. This record does not assert independent reproduction.

### Qwen2.5-VL Agent

Action Accuracy: 70.4. Result ID: 6a3b969cfbc373592c44b22e.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773a1e209131cea71d3c.md>)

Recorded evaluation configuration: {"base\_agent":"Qwen2.5-VL Agent"}. This record does not assert independent reproduction.

### ShowUI + VisCritic

Action Accuracy: 69.8. Result ID: 6a3b969cfbc373592c44b22b.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

Recorded evaluation configuration: {"base\_agent":"ShowUI","critic\_module":"VisCritic"}. This record does not assert independent reproduction.

### ShowUI + GUI-PRA

Action Accuracy: 68.8. Result ID: 6a3b969cfbc373592c44b22a.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

Recorded evaluation configuration: {"base\_agent":"ShowUI","critic\_module":"GUI-PRA"}. This record does not assert independent reproduction.

### CogAgent + VisCritic

Action Accuracy: 67.1. Result ID: 6a3b969cfbc373592c44b22d.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

Recorded evaluation configuration: {"base\_agent":"CogAgent","critic\_module":"VisCritic"}. This record does not assert independent reproduction.

### ShowUI

Action Accuracy: 66.7. Result ID: 6a3b969cfbc373592c44b229.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75e41e209131cea6a7f9.md>)

[Implementation](<https://github.com/showlab/ShowUI>)

Recorded evaluation configuration: {"base\_agent":"ShowUI"}. This record does not assert independent reproduction.

### SeeClick + GUI-Critic-R1

Action Accuracy: 64.8. Result ID: 6a3b969cfbc373592c44b227.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

Recorded evaluation configuration: {"base\_agent":"SeeClick","critic\_module":"GUI-Critic-R1"}. This record does not assert independent reproduction.

### CogAgent

Action Accuracy: 64.8. Result ID: 6a3b969cfbc373592c44b22c.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76711e209131cea6d06f.md>)

[Implementation](<https://github.com/THUDM/CogVLM>)

Recorded evaluation configuration: {"base\_agent":"CogAgent"}. This record does not assert independent reproduction.

### SeeClick + GUI-PRA

Action Accuracy: 64.5. Result ID: 6a3b969cfbc373592c44b225.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

Recorded evaluation configuration: {"base\_agent":"SeeClick","critic\_module":"GUI-PRA"}. This record does not assert independent reproduction.

### SeeClick + VisCritic

Action Accuracy: 64.5. Result ID: 6a3b969cfbc373592c44b228.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

Recorded evaluation configuration: {"base\_agent":"SeeClick","critic\_module":"VisCritic"}. This record does not assert independent reproduction.

### SeeClick + BacktrackAgent

Action Accuracy: 64.2. Result ID: 6a3b969cfbc373592c44b226.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

Recorded evaluation configuration: {"base\_agent":"SeeClick","critic\_module":"BacktrackAgent"}. This record does not assert independent reproduction.

### SeeClick

Action Accuracy: 62.1. Result ID: 6a3b969cfbc373592c44b224.

[Reporting paper](<https://www.sota2.com/research/paper/6a3b64d8206542251d7c993a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72441e209131cea55627.md>)

[Implementation](<https://github.com/njucckevin/SeeClick>)

Recorded evaluation configuration: {"base\_agent":"SeeClick"}. This record does not assert independent reproduction.

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