# Optical Character Recognition on OCRBench English v2

Source: [Optical Character Recognition on OCRBench English v2](<https://www.sota2.com/research/sota/optical-character-recognition-on-ocrbench-english-v2>)

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

Data updated: 2026-02-26T06:01:43.881Z

Evaluation of Optical Character Recognition performance on the OCRBench English v2 dataset, reporting detailed sub-task scores including Recognition, Extraction, Understanding, and the overall Average.

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 | Optical Character Recognition |
| Dataset (including any stated split) | OCRBench English v2 |
| Metric | Recognition |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 19 of 19 |
| 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

- [Recognition](<https://www.sota2.com/research/sota/optical-character-recognition-on-ocrbench-english-v2.md?metric=Recognition>): Higher is better; unit: PERCENT.

- [Referring](<https://www.sota2.com/research/sota/optical-character-recognition-on-ocrbench-english-v2.md?metric=Referring>): Higher is better; unit: PERCENT.

- [Spotting](<https://www.sota2.com/research/sota/optical-character-recognition-on-ocrbench-english-v2.md?metric=Spotting>): Higher is better; unit: PERCENT.

- [Extraction](<https://www.sota2.com/research/sota/optical-character-recognition-on-ocrbench-english-v2.md?metric=Extraction>): Higher is better; unit: PERCENT.

- [Parsing](<https://www.sota2.com/research/sota/optical-character-recognition-on-ocrbench-english-v2.md?metric=Parsing>): Higher is better; unit: PERCENT.

- [Calculation](<https://www.sota2.com/research/sota/optical-character-recognition-on-ocrbench-english-v2.md?metric=Calculation>): Higher is better; unit: PERCENT.

- [Understanding](<https://www.sota2.com/research/sota/optical-character-recognition-on-ocrbench-english-v2.md?metric=Understanding>): Higher is better; unit: PERCENT.

- [Reasoning](<https://www.sota2.com/research/sota/optical-character-recognition-on-ocrbench-english-v2.md?metric=Reasoning>): Higher is better; unit: PERCENT.

- [Average Score](<https://www.sota2.com/research/sota/optical-character-recognition-on-ocrbench-english-v2.md?metric=Average%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Recognition | Result date | Report paper |
| --- | --- | --- | --- |
| Ovis2-8B | 73.2 | Not recorded | 699e82176017c4460ffacfde |
| OCR-Agent-7B | 71.8 | Not recorded | 699e82176017c4460ffacfde |
| InternVL3-8B | 68.6 | Not recorded | 699e82176017c4460ffacfde |
| Step-1V | 67.8 | Not recorded | 699e82176017c4460ffacfde |
| MiniCPM-V-2.6-7B | 66.9 | Not recorded | 699e82176017c4460ffacfde |
| RolmOCR-7B | 63.3 | Not recorded | 699e82176017c4460ffacfde |
| Deepseek-vl-2-Small-16B | 62.7 | Not recorded | 699e82176017c4460ffacfde |
| Claude3.5-sonnet | 62.2 | Not recorded | 699e82176017c4460ffacfde |
| GLM-4V-9B | 61.8 | Not recorded | 699e82176017c4460ffacfde |
| GPT-4o | 61.2 | Not recorded | 699e82176017c4460ffacfde |
| Gemini-Pro | 61.2 | Not recorded | 699e82176017c4460ffacfde |
| GPT-4o-mini | 57.9 | Not recorded | 699e82176017c4460ffacfde |
| Molmo-7B | 52.4 | Not recorded | 699e82176017c4460ffacfde |
| Pixtran-12B | 48.9 | Not recorded | 699e82176017c4460ffacfde |
| LLaVA-OV-7B | 46 | Not recorded | 699e82176017c4460ffacfde |
| Cambrian-1-8B | 45.3 | Not recorded | 699e82176017c4460ffacfde |
| LLaVA-Next-8B | 41.3 | Not recorded | 699e82176017c4460ffacfde |
| TextMonkey-8B | 39.1 | Not recorded | 699e82176017c4460ffacfde |
| Monkey-8B | 35.2 | Not recorded | 699e82176017c4460ffacfde |

### Ovis2-8B

Recognition: 73.2. Result ID: 699ebee45698665ec48e6e92.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Open-source VLMs"}. This record does not assert independent reproduction.

### OCR-Agent-7B

Recognition: 71.8. Result ID: 699ebee45698665ec48e6e99.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

[Method origin paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

[Implementation](<https://github.com/AIGeeksGroup/OCR-Agent>)

Recorded evaluation configuration: {"Model Category":"Ours"}. This record does not assert independent reproduction.

### InternVL3-8B

Recognition: 68.6. Result ID: 699ebee45698665ec48e6e8e.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Open-source VLMs"}. This record does not assert independent reproduction.

### Step-1V

Recognition: 67.8. Result ID: 699ebee45698665ec48e6e98.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Closed-source VLMs"}. This record does not assert independent reproduction.

### MiniCPM-V-2.6-7B

Recognition: 66.9. Result ID: 699ebee45698665ec48e6e90.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Open-source VLMs"}. This record does not assert independent reproduction.

### RolmOCR-7B

Recognition: 63.3. Result ID: 699ebee45698665ec48e6e93.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Open-source VLMs"}. This record does not assert independent reproduction.

### Deepseek-vl-2-Small-16B

Recognition: 62.7. Result ID: 699ebee45698665ec48e6e8f.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Open-source VLMs"}. This record does not assert independent reproduction.

### Claude3.5-sonnet

Recognition: 62.2. Result ID: 699ebee45698665ec48e6e97.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Closed-source VLMs"}. This record does not assert independent reproduction.

### GLM-4V-9B

Recognition: 61.8. Result ID: 699ebee45698665ec48e6e91.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Open-source VLMs"}. This record does not assert independent reproduction.

### GPT-4o

Recognition: 61.2. Result ID: 699ebee45698665ec48e6e94.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Closed-source VLMs"}. This record does not assert independent reproduction.

### Gemini-Pro

Recognition: 61.2. Result ID: 699ebee45698665ec48e6e96.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Closed-source VLMs"}. This record does not assert independent reproduction.

### GPT-4o-mini

Recognition: 57.9. Result ID: 699ebee45698665ec48e6e95.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Closed-source VLMs"}. This record does not assert independent reproduction.

### Molmo-7B

Recognition: 52.4. Result ID: 699ebee45698665ec48e6e8b.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Open-source VLMs"}. This record does not assert independent reproduction.

### Pixtran-12B

Recognition: 48.9. Result ID: 699ebee45698665ec48e6e8d.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Open-source VLMs"}. This record does not assert independent reproduction.

### LLaVA-OV-7B

Recognition: 46. Result ID: 699ebee45698665ec48e6e88.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Open-source VLMs"}. This record does not assert independent reproduction.

### Cambrian-1-8B

Recognition: 45.3. Result ID: 699ebee45698665ec48e6e8c.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Open-source VLMs"}. This record does not assert independent reproduction.

### LLaVA-Next-8B

Recognition: 41.3. Result ID: 699ebee45698665ec48e6e87.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Open-source VLMs"}. This record does not assert independent reproduction.

### TextMonkey-8B

Recognition: 39.1. Result ID: 699ebee45698665ec48e6e8a.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Open-source VLMs"}. This record does not assert independent reproduction.

### Monkey-8B

Recognition: 35.2. Result ID: 699ebee45698665ec48e6e89.

[Reporting paper](<https://www.sota2.com/research/paper/699e82176017c4460ffacfde.md>)

Recorded evaluation configuration: {"Model Category":"Open-source VLMs"}. This record does not assert independent reproduction.

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