# Text-to-Image on GenExam Math (Overall)

Source: [Text-to-Image on GenExam Math (Overall)](<https://www.sota2.com/research/sota/text-to-image-on-genexam-math-overall>)

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

Data updated: 2026-05-21T11:45:10.020Z

Evaluation of text-to-image generation models specifically on the Math subset of the GenExam dataset, reporting Strict and Relaxed accuracy 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 | text-to-image |
| Dataset (including any stated split) | GenExam Math (Overall) |
| Metric | Strict Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 14 of 14 |
| 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

- [Strict Accuracy](<https://www.sota2.com/research/sota/text-to-image-on-genexam-math-overall.md?metric=Strict%20Accuracy>): Higher is better; unit: PERCENT.

- [Relaxed Accuracy](<https://www.sota2.com/research/sota/text-to-image-on-genexam-math-overall.md?metric=Relaxed%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Strict Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Draw2Think | 68.2 | Not recorded | 6a0e916ff984739ed6738980 |
| Nano Banana 2 | 56.3 | Not recorded | 6a0e916ff984739ed6738980 |
| Nano Banana Pro | 55.6 | Not recorded | 6a0e916ff984739ed6738980 |
| GPT-Image-2 | 50.3 | Not recorded | 6a0e916ff984739ed6738980 |
| Seedream 5.0 | 47 | Not recorded | 6a0e916ff984739ed6738980 |
| GPT-Image-1.5 | 26.5 | Not recorded | 6a0e916ff984739ed6738980 |
| Faire | 9.3 | Not recorded | 6a0e916ff984739ed6738980 |
| FLUX.2 max | 6.6 | Not recorded | 6a0e916ff984739ed6738980 |
| FLUX.2 dev | 2.6 | Not recorded | 6a0e916ff984739ed6738980 |
| Qwen-Image-2512 | 0 | Not recorded | 6a0e916ff984739ed6738980 |
| HunyuanImage-3.0 | 0 | Not recorded | 6a0e916ff984739ed6738980 |
| BLIP3o-NEXT-GRPO | 0 | Not recorded | 6a0e916ff984739ed6738980 |
| BAGEL | 0 | Not recorded | 6a0e916ff984739ed6738980 |
| Janus-Pro | 0 | Not recorded | 6a0e916ff984739ed6738980 |

### Draw2Think

Strict Accuracy: 68.2. Result ID: 6a0eef90fbc373592c073b38.

[Reporting paper](<https://www.sota2.com/research/paper/6a0e916ff984739ed6738980.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0e916ff984739ed6738980.md>)

Recorded evaluation configuration: {"Model Category":"Closed-source Models","Base Model / Release Information":"Gemini 3 Flash Preview \| Dec. 2025"}. This record does not assert independent reproduction.

### Nano Banana 2

Strict Accuracy: 56.3. Result ID: 6a0eef90fbc373592c073b39.

[Reporting paper](<https://www.sota2.com/research/paper/6a0e916ff984739ed6738980.md>)

Recorded evaluation configuration: {"Model Category":"Closed-source Models","Base Model / Release Information":"Gemini 3.1 Flash Image \| Feb. 2026"}. This record does not assert independent reproduction.

### Nano Banana Pro

Strict Accuracy: 55.6. Result ID: 6a0eef90fbc373592c073b3a.

[Reporting paper](<https://www.sota2.com/research/paper/6a0e916ff984739ed6738980.md>)

Recorded evaluation configuration: {"Model Category":"Closed-source Models","Base Model / Release Information":"Gemini 3 Pro Image \| Nov. 2025"}. This record does not assert independent reproduction.

### GPT-Image-2

Strict Accuracy: 50.3. Result ID: 6a0eef90fbc373592c073b3b.

[Reporting paper](<https://www.sota2.com/research/paper/6a0e916ff984739ed6738980.md>)

Recorded evaluation configuration: {"Model Category":"Closed-source Models","Base Model / Release Information":"OpenAI native T2I \| Apr. 2026"}. This record does not assert independent reproduction.

### Seedream 5.0

Strict Accuracy: 47. Result ID: 6a0eef90fbc373592c073b3c.

[Reporting paper](<https://www.sota2.com/research/paper/6a0e916ff984739ed6738980.md>)

Recorded evaluation configuration: {"Model Category":"Closed-source Models","Base Model / Release Information":"ByteDance native T2I \| Feb. 2026"}. This record does not assert independent reproduction.

### GPT-Image-1.5

Strict Accuracy: 26.5. Result ID: 6a0eef90fbc373592c073b3d.

[Reporting paper](<https://www.sota2.com/research/paper/6a0e916ff984739ed6738980.md>)

Recorded evaluation configuration: {"Model Category":"Closed-source Models","Base Model / Release Information":"OpenAI native T2I \| Dec. 2025"}. This record does not assert independent reproduction.

### Faire

Strict Accuracy: 9.3. Result ID: 6a0eef90fbc373592c073b45.

[Reporting paper](<https://www.sota2.com/research/paper/6a0e916ff984739ed6738980.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69a66c132d603e6de3e3cd8b.md>)

Recorded evaluation configuration: {"Model Category":"Open-source Code-rendered RL","Base Model / Release Information":"Qwen3-VL-8B + GRPO"}. This record does not assert independent reproduction.

### FLUX.2 max

Strict Accuracy: 6.6. Result ID: 6a0eef90fbc373592c073b3e.

[Reporting paper](<https://www.sota2.com/research/paper/6a0e916ff984739ed6738980.md>)

Recorded evaluation configuration: {"Model Category":"Closed-source Models","Base Model / Release Information":"Black Forest Labs T2I \| Dec. 2025"}. This record does not assert independent reproduction.

### FLUX.2 dev

Strict Accuracy: 2.6. Result ID: 6a0eef90fbc373592c073b3f.

[Reporting paper](<https://www.sota2.com/research/paper/6a0e916ff984739ed6738980.md>)

Recorded evaluation configuration: {"Model Category":"Open-source T2I Models"}. This record does not assert independent reproduction.

### Qwen-Image-2512

Strict Accuracy: 0. Result ID: 6a0eef90fbc373592c073b40.

[Reporting paper](<https://www.sota2.com/research/paper/6a0e916ff984739ed6738980.md>)

Recorded evaluation configuration: {"Model Category":"Open-source T2I Models"}. This record does not assert independent reproduction.

### HunyuanImage-3.0

Strict Accuracy: 0. Result ID: 6a0eef90fbc373592c073b41.

[Reporting paper](<https://www.sota2.com/research/paper/6a0e916ff984739ed6738980.md>)

Recorded evaluation configuration: {"Model Category":"Open-source T2I Models"}. This record does not assert independent reproduction.

### BLIP3o-NEXT-GRPO

Strict Accuracy: 0. Result ID: 6a0eef90fbc373592c073b42.

[Reporting paper](<https://www.sota2.com/research/paper/6a0e916ff984739ed6738980.md>)

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

### BAGEL

Strict Accuracy: 0. Result ID: 6a0eef90fbc373592c073b43.

[Reporting paper](<https://www.sota2.com/research/paper/6a0e916ff984739ed6738980.md>)

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

### Janus-Pro

Strict Accuracy: 0. Result ID: 6a0eef90fbc373592c073b44.

[Reporting paper](<https://www.sota2.com/research/paper/6a0e916ff984739ed6738980.md>)

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

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