# Visual Perception on HR-8K (test)

Source: [Visual Perception on HR-8K (test)](<https://www.sota2.com/research/sota/visual-perception-on-hr-8k-test>)

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

Data updated: 2026-02-26T05:31:20.814Z

Evaluation of model performance on the test split of the HR-8K dataset, focusing on Visual Perception capabilities measured by 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 | Visual Perception |
| Dataset (including any stated split) | HR-8K (test) |
| Metric | 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

- [Accuracy](<https://www.sota2.com/research/sota/visual-perception-on-hr-8k-test.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| ZwZ-8B | 82 | Not recorded | 698eb03ec97a0b3e3b5e6800 |
| Skywork-R1V4 | 79.8 | Not recorded | 698eb03ec97a0b3e3b5e6800 |
| ZwZ-4B | 79.5 | Not recorded | 698eb03ec97a0b3e3b5e6800 |
| SenseNova-MARS-8B | 78.4 | Not recorded | 698eb03ec97a0b3e3b5e6800 |
| Qwen3-VL-8B+tool | 78 | Not recorded | 698eb03ec97a0b3e3b5e6800 |
| Qwen3-VL-8B | 74.6 | Not recorded | 698eb03ec97a0b3e3b5e6800 |
| Qwen3-VL-4B+tool | 74.4 | Not recorded | 698eb03ec97a0b3e3b5e6800 |
| DeepEyesV2 | 73.8 | Not recorded | 698eb03ec97a0b3e3b5e6800 |
| Mini o3 | 73.3 | Not recorded | 698eb03ec97a0b3e3b5e6800 |
| ZwZ-7B | 73.3 | Not recorded | 698eb03ec97a0b3e3b5e6800 |
| Qwen3-VL-4B | 72.9 | Not recorded | 698eb03ec97a0b3e3b5e6800 |
| DeepEyes | 72.6 | Not recorded | 698eb03ec97a0b3e3b5e6800 |
| Thyme | 72 | Not recorded | 698eb03ec97a0b3e3b5e6800 |
| Qwen2.5-VL-7B | 67.9 | Not recorded | 698eb03ec97a0b3e3b5e6800 |
| Pixel-Reasoner | 66.1 | Not recorded | 698eb03ec97a0b3e3b5e6800 |

### ZwZ-8B

Accuracy: 82. Result ID: 69970dfa5698665ec45b7a88.

[Reporting paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Implementation](<https://github.com/inclusionAI/Zooming-without-Zooming>)

Recorded evaluation configuration: {"Inference Strategy":"Single Forward Pass","Model Scale":"8B"}. This record does not assert independent reproduction.

### Skywork-R1V4

Accuracy: 79.8. Result ID: 69970dfa5698665ec45b7a83.

[Reporting paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Method origin paper](<https://www.sota2.com/research/paper/692fc402d6b0dd3342933253.md>)

[Implementation](<https://github.com/SkyworkAI/Skywork-R1V/tree/main/r1v4>)

Recorded evaluation configuration: {"Inference Strategy":"Agentic","Model Scale":"30B"}. This record does not assert independent reproduction.

### ZwZ-4B

Accuracy: 79.5. Result ID: 69970dfa5698665ec45b7a86.

[Reporting paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Implementation](<https://github.com/inclusionAI/Zooming-without-Zooming>)

Recorded evaluation configuration: {"Inference Strategy":"Single Forward Pass","Model Scale":"4B"}. This record does not assert independent reproduction.

### SenseNova-MARS-8B

Accuracy: 78.4. Result ID: 69970dfa5698665ec45b7a82.

[Reporting paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6955ff6653fdd5c36173de38.md>)

[Implementation](<https://github.com/OpenSenseNova/SenseNova-MARS>)

Recorded evaluation configuration: {"Inference Strategy":"Agentic","Model Scale":"8B"}. This record does not assert independent reproduction.

### Qwen3-VL-8B+tool

Accuracy: 78. Result ID: 69970dfa5698665ec45b7a85.

[Reporting paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6927db54d6b0dd3342930f66.md>)

Recorded evaluation configuration: {"Inference Strategy":"Tool Use","Model Scale":"8B"}. This record does not assert independent reproduction.

### Qwen3-VL-8B

Accuracy: 74.6. Result ID: 69970dfa5698665ec45b7a7c.

[Reporting paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6927db54d6b0dd3342930f66.md>)

Recorded evaluation configuration: {"Inference Strategy":"Single Forward Pass","Model Scale":"8B"}. This record does not assert independent reproduction.

### Qwen3-VL-4B+tool

Accuracy: 74.4. Result ID: 69970dfa5698665ec45b7a84.

[Reporting paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6927db54d6b0dd3342930f66.md>)

Recorded evaluation configuration: {"Inference Strategy":"Tool Use","Model Scale":"4B"}. This record does not assert independent reproduction.

### DeepEyesV2

Accuracy: 73.8. Result ID: 69970dfa5698665ec45b7a80.

[Reporting paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Method origin paper](<https://www.sota2.com/research/paper/691171652828b7cc6937d492.md>)

[Implementation](<https://github.com/Visual-Agent/DeepEyes>)

Recorded evaluation configuration: {"Inference Strategy":"Agentic","Model Scale":"7B"}. This record does not assert independent reproduction.

### Mini o3

Accuracy: 73.3. Result ID: 69970dfa5698665ec45b7a81.

[Reporting paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b2c1e209131cea87068.md>)

Recorded evaluation configuration: {"Inference Strategy":"Agentic","Model Scale":"7B"}. This record does not assert independent reproduction.

### ZwZ-7B

Accuracy: 73.3. Result ID: 69970dfa5698665ec45b7a87.

[Reporting paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Implementation](<https://github.com/inclusionAI/Zooming-without-Zooming>)

Recorded evaluation configuration: {"Inference Strategy":"Single Forward Pass","Model Scale":"7B"}. This record does not assert independent reproduction.

### Qwen3-VL-4B

Accuracy: 72.9. Result ID: 69970dfa5698665ec45b7a7a.

[Reporting paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6927db54d6b0dd3342930f66.md>)

Recorded evaluation configuration: {"Inference Strategy":"Single Forward Pass","Model Scale":"4B"}. This record does not assert independent reproduction.

### DeepEyes

Accuracy: 72.6. Result ID: 69970dfa5698665ec45b7a7e.

[Reporting paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78c71e209131cea7a75e.md>)

Recorded evaluation configuration: {"Inference Strategy":"Agentic","Model Scale":"7B"}. This record does not assert independent reproduction.

### Thyme

Accuracy: 72. Result ID: 69970dfa5698665ec45b7a7f.

[Reporting paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7aa21e209131cea84434.md>)

Recorded evaluation configuration: {"Inference Strategy":"Agentic","Model Scale":"7B"}. This record does not assert independent reproduction.

### Qwen2.5-VL-7B

Accuracy: 67.9. Result ID: 69970dfa5698665ec45b7a7b.

[Reporting paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

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

Recorded evaluation configuration: {"Inference Strategy":"Single Forward Pass","Model Scale":"7B"}. This record does not assert independent reproduction.

### Pixel-Reasoner

Accuracy: 66.1. Result ID: 69970dfa5698665ec45b7a7d.

[Reporting paper](<https://www.sota2.com/research/paper/698eb03ec97a0b3e3b5e6800.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78c71e209131cea7a774.md>)

Recorded evaluation configuration: {"Inference Strategy":"Agentic","Model Scale":"7B"}. This record does not assert independent reproduction.

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