# Text-to-Image Generation on CompBench (Numeracy, 3D Spatial, Overall)

Source: [Text-to-Image Generation on CompBench (Numeracy, 3D Spatial, Overall)](<https://www.sota2.com/research/sota/text-to-image-generation-on-compbench-numeracy-3d-spatial-overall>)

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

Data updated: 2026-03-23T11:26:26.276Z

Evaluation of Text-to-Image generation models on the CompBench dataset, focusing on performance across Numeracy, 3D Spatial reasoning, and the Overall score.

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 Generation |
| Dataset (including any stated split) | CompBench |
| Metric | Overall Score |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 34 of 34 |
| 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

- [Overall Score](<https://www.sota2.com/research/sota/text-to-image-generation-on-compbench-numeracy-3d-spatial-overall.md?metric=Overall%20Score>): Higher is better; unit: PERCENT.

- [Numeracy](<https://www.sota2.com/research/sota/text-to-image-generation-on-compbench-numeracy-3d-spatial-overall.md?metric=Numeracy>): Higher is better; unit: PERCENT.

- [3D Spatial](<https://www.sota2.com/research/sota/text-to-image-generation-on-compbench-numeracy-3d-spatial-overall.md?metric=3D%20Spatial>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Overall Score | Result date | Report paper |
| --- | --- | --- | --- |
| UniCorn | 88.5 | Not recorded | 695de86d53fdd5c361743b90 |
| SeGroS | 88.08 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| SFT\* | 87.32 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| Reca | 87.2 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| SeGroS | 86.9 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| SeGroS | 86.14 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| Reca | 86 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| OmniGen2 | 85.8 | Not recorded | 695de86d53fdd5c361743b90 |
| Reca | 85.7 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| BLIP3-o | 84.7 | Not recorded | 695de86d53fdd5c361743b90 |
| Reca | 84.47 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| SeGroS | 84.35 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| SD3 Medium | 84.3 | Not recorded | 695de86d53fdd5c361743b90 |
| SeGroS | 83.83 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| SFT\* | 83.5 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| FLUX.1 dev | 83.1 | Not recorded | 695de86d53fdd5c361743b90 |
| Reca | 83.1 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| show-o2 | 82.8 | Not recorded | 695de86d53fdd5c361743b90 |
| SFT\* | 82.6 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| BAGEL | 82.2 | Not recorded | 695de86d53fdd5c361743b90 |
| T2I-R1 | 81.9 | Not recorded | 695de86d53fdd5c361743b90 |
| w/o SFT | 81.36 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| SFT\* | 81.25 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| w/o SFT | 80.53 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| w/o SFT | 80.34 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| SFT\* | 80.02 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| w/o SFT | 78.84 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| SeGroS | 78.3 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| Reca | 77.5 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| w/o SFT | 75.47 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| SFT\* | 75.37 | Not recorded | 69c0c8e30636e5bf4afc39f6 |
| Janus-Pro | 74 | Not recorded | 695de86d53fdd5c361743b90 |
| w/o SFT | 73.58 | Not recorded | 69c0c8e30636e5bf4afc39f6 |

### UniCorn

Overall Score: 88.5. Result ID: 6997e15f5698665ec43a3469.

[Reporting paper](<https://www.sota2.com/research/paper/695de86d53fdd5c361743b90.md>)

[Method origin paper](<https://www.sota2.com/research/paper/695de86d53fdd5c361743b90.md>)

[Implementation](<https://github.com/Hungryyan1/UniCorn>)

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

### SeGroS

Overall Score: 88.08. Result ID: 69c1235947f34ce924a9973b.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"Harmon-1.5B","Fine-Tuning Dataset":"BLIP3o-60k Dataset"}. This record does not assert independent reproduction.

### SFT\*

Overall Score: 87.32. Result ID: 69c1235947f34ce924a99739.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"Harmon-1.5B","Fine-Tuning Dataset":"BLIP3o-60k Dataset"}. This record does not assert independent reproduction.

### Reca

Overall Score: 87.2. Result ID: 69c1235947f34ce924a9973a.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b601e209131cea88021.md>)

Recorded evaluation configuration: {"UMM":"Harmon-1.5B","Fine-Tuning Dataset":"BLIP3o-60k Dataset"}. This record does not assert independent reproduction.

### SeGroS

Overall Score: 86.9. Result ID: 69c1235947f34ce924a99737.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"Harmon-0.5B","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### SeGroS

Overall Score: 86.14. Result ID: 69c1235947f34ce924a99733.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"OpenUni-3.6B","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### Reca

Overall Score: 86. Result ID: 69c1235947f34ce924a99732.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b601e209131cea88021.md>)

Recorded evaluation configuration: {"UMM":"OpenUni-3.6B","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### OmniGen2

Overall Score: 85.8. Result ID: 6997e15f5698665ec43a3465.

[Reporting paper](<https://www.sota2.com/research/paper/695de86d53fdd5c361743b90.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75d81e209131cea6a322.md>)

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

### Reca

Overall Score: 85.7. Result ID: 69c1235947f34ce924a99736.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b601e209131cea88021.md>)

Recorded evaluation configuration: {"UMM":"Harmon-0.5B","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### BLIP3-o

Overall Score: 84.7. Result ID: 6997e15f5698665ec43a3464.

[Reporting paper](<https://www.sota2.com/research/paper/695de86d53fdd5c361743b90.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78841e209131cea78c86.md>)

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

### Reca

Overall Score: 84.47. Result ID: 69c1235947f34ce924a9972a.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b601e209131cea88021.md>)

Recorded evaluation configuration: {"UMM":"Show-o-512","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### SeGroS

Overall Score: 84.35. Result ID: 69c1235947f34ce924a9972b.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"Show-o-512","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### SD3 Medium

Overall Score: 84.3. Result ID: 6997e15f5698665ec43a3460.

[Reporting paper](<https://www.sota2.com/research/paper/695de86d53fdd5c361743b90.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72621e209131cea56163.md>)

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

### SeGroS

Overall Score: 83.83. Result ID: 69c1235947f34ce924a9972f.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"OpenUni-1.6B","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### SFT\*

Overall Score: 83.5. Result ID: 69c1235947f34ce924a99735.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"Harmon-0.5B","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### FLUX.1 dev

Overall Score: 83.1. Result ID: 6997e15f5698665ec43a3461.

[Reporting paper](<https://www.sota2.com/research/paper/695de86d53fdd5c361743b90.md>)

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

### Reca

Overall Score: 83.1. Result ID: 69c1235947f34ce924a9972e.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b601e209131cea88021.md>)

Recorded evaluation configuration: {"UMM":"OpenUni-1.6B","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### show-o2

Overall Score: 82.8. Result ID: 6997e15f5698665ec43a3463.

[Reporting paper](<https://www.sota2.com/research/paper/695de86d53fdd5c361743b90.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b761e209131cea88a1f.md>)

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

### SFT\*

Overall Score: 82.6. Result ID: 69c1235947f34ce924a99731.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"OpenUni-3.6B","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### BAGEL

Overall Score: 82.2. Result ID: 6997e15f5698665ec43a3468.

[Reporting paper](<https://www.sota2.com/research/paper/695de86d53fdd5c361743b90.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a311e209131cea81aad.md>)

[Implementation](<https://bagel-ai.org/>)

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

### T2I-R1

Overall Score: 81.9. Result ID: 6997e15f5698665ec43a3467.

[Reporting paper](<https://www.sota2.com/research/paper/695de86d53fdd5c361743b90.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79cc1e209131cea7f6a9.md>)

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

### w/o SFT

Overall Score: 81.36. Result ID: 69c1235947f34ce924a99738.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"Harmon-1.5B","Fine-Tuning Dataset":"BLIP3o-60k Dataset"}. This record does not assert independent reproduction.

### SFT\*

Overall Score: 81.25. Result ID: 69c1235947f34ce924a9972d.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"OpenUni-1.6B","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### w/o SFT

Overall Score: 80.53. Result ID: 69c1235947f34ce924a99728.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"Show-o-512","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### w/o SFT

Overall Score: 80.34. Result ID: 69c1235947f34ce924a99734.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"Harmon-0.5B","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### SFT\*

Overall Score: 80.02. Result ID: 69c1235947f34ce924a99729.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"Show-o-512","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### w/o SFT

Overall Score: 78.84. Result ID: 69c1235947f34ce924a99730.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"OpenUni-3.6B","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### SeGroS

Overall Score: 78.3. Result ID: 69c1235947f34ce924a99727.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"Show-o-256","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### Reca

Overall Score: 77.5. Result ID: 69c1235947f34ce924a99726.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b601e209131cea88021.md>)

Recorded evaluation configuration: {"UMM":"Show-o-256","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### w/o SFT

Overall Score: 75.47. Result ID: 69c1235947f34ce924a9972c.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"OpenUni-1.6B","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### SFT\*

Overall Score: 75.37. Result ID: 69c1235947f34ce924a99725.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"Show-o-256","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

### Janus-Pro

Overall Score: 74. Result ID: 6997e15f5698665ec43a3462.

[Reporting paper](<https://www.sota2.com/research/paper/695de86d53fdd5c361743b90.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76cc1e209131cea6f20a.md>)

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

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

### w/o SFT

Overall Score: 73.58. Result ID: 69c1235947f34ce924a99724.

[Reporting paper](<https://www.sota2.com/research/paper/69c0c8e30636e5bf4afc39f6.md>)

Recorded evaluation configuration: {"UMM":"Show-o-256","Fine-Tuning Dataset":"MidjourneyV6 Dataset"}. This record does not assert independent reproduction.

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