# Planning and Puzzle Solving on Sokoban

Source: [Planning and Puzzle Solving on Sokoban](<https://www.sota2.com/research/sota/planning-and-puzzle-solving-on-sokoban>)

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

Data updated: 2026-04-14T09:56:14.504Z

Evaluation of model performance on solving Sokoban puzzles using Accuracy (Acc.) and Completion (Comp.) 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 | Planning and puzzle solving |
| Dataset (including any stated split) | Sokoban |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| 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/planning-and-puzzle-solving-on-sokoban.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Completion](<https://www.sota2.com/research/sota/planning-and-puzzle-solving-on-sokoban.md?metric=Completion>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Skill-SD | 62.5 | Not recorded | 69ddca1b7d0172140ec79a5f |
| Vanilla GRPO | 51.6 | Not recorded | 69ddca1b7d0172140ec79a5f |
| Vanilla OPD | 21.9 | Not recorded | 69ddca1b7d0172140ec79a5f |
| Skill-Augmented GRPO | 20.3 | Not recorded | 69ddca1b7d0172140ec79a5f |
| Base Model | 12.5 | Not recorded | 69ddca1b7d0172140ec79a5f |

### Skill-SD

Accuracy: 62.5. Result ID: 69de0f3647f34ce924c5f3f0.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca1b7d0172140ec79a5f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ddca1b7d0172140ec79a5f.md>)

Recorded evaluation configuration: {"Model Backbone":"Qwen3-4B-Instruct-2507"}. This record does not assert independent reproduction.

### Vanilla GRPO

Accuracy: 51.6. Result ID: 69de0f3647f34ce924c5f3ee.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca1b7d0172140ec79a5f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73101e209131cea5a395.md>)

[Implementation](<https://github.com/deepseek-ai/DeepSeek-Math>)

Recorded evaluation configuration: {"Model Backbone":"Qwen3-4B-Instruct-2507"}. This record does not assert independent reproduction.

### Vanilla OPD

Accuracy: 21.9. Result ID: 69de0f3647f34ce924c5f3ed.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca1b7d0172140ec79a5f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71dd1e209131cea536d1.md>)

Recorded evaluation configuration: {"Model Backbone":"Qwen3-4B-Instruct-2507"}. This record does not assert independent reproduction.

### Skill-Augmented GRPO

Accuracy: 20.3. Result ID: 69de0f3647f34ce924c5f3ef.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca1b7d0172140ec79a5f.md>)

Recorded evaluation configuration: {"Model Backbone":"Qwen3-4B-Instruct-2507"}. This record does not assert independent reproduction.

### Base Model

Accuracy: 12.5. Result ID: 69de0f3647f34ce924c5f3ec.

[Reporting paper](<https://www.sota2.com/research/paper/69ddca1b7d0172140ec79a5f.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa788e1e209131cea79116.md>)

Recorded evaluation configuration: {"Model Backbone":"Qwen3-4B-Instruct-2507"}. This record does not assert independent reproduction.

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