# Text-to-Image Generation on GenEval original prompts LLM-rewritten prompts Source: [Text-to-Image Generation on GenEval original prompts LLM-rewritten prompts]() Collection index: [llms.txt]() Data updated: 2026-05-19T09:11:48.782Z Evaluation of Text-to-Image generation models using a specific variant of the GenEval dataset where the original prompts have been rewritten by an LLM. Evaluation focuses on compositional metrics like single object accuracy, counting, and attribute binding. 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) | GenEval original prompts LLM-rewritten prompts | | Metric | Overall Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 17 of 17 | | 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 Accuracy](): Higher is better; unit: PERCENT. - [Single Object Accuracy](): Higher is better; unit: PERCENT. - [Two Object Accuracy](): Higher is better; unit: PERCENT. - [Counting Accuracy](): Higher is better; unit: PERCENT. - [Color Accuracy](): Higher is better; unit: PERCENT. - [Position Accuracy](): Higher is better; unit: PERCENT. - [Color Attribute Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | Overall Accuracy | Result date | Report paper | | --- | --- | --- | --- | | LAC | 90 | Not recorded | 6a0beeb9dbc836b87d82c593 | | Mogao | 89 | Not recorded | 6a0beeb9dbc836b87d82c593 | | BAGEL | 86 | Not recorded | 6a0beeb9dbc836b87d82c593 | | FLUX.1-dev | 82 | Not recorded | 6a0beeb9dbc836b87d82c593 | | LAC | 82 | Not recorded | 6a0beeb9dbc836b87d82c593 | | MetaQuery-XL | 80 | Not recorded | 6a0beeb9dbc836b87d82c593 | | Janus-Pro-7B | 80 | Not recorded | 6a0beeb9dbc836b87d82c593 | | BAGEL | 77 | Not recorded | 6a0beeb9dbc836b87d82c593 | | Show-o2 | 76 | Not recorded | 6a0beeb9dbc836b87d82c593 | | SD3-Medium | 74 | Not recorded | 6a0beeb9dbc836b87d82c593 | | DALL-E 3 | 67 | Not recorded | 6a0beeb9dbc836b87d82c593 | | Emu3-Gen | 66 | Not recorded | 6a0beeb9dbc836b87d82c593 | | Transfusion | 63 | Not recorded | 6a0beeb9dbc836b87d82c593 | | TokenFlow-XL | 55 | Not recorded | 6a0beeb9dbc836b87d82c593 | | SEED-X | 49 | Not recorded | 6a0beeb9dbc836b87d82c593 | | PixArt-Σ | 48 | Not recorded | 6a0beeb9dbc836b87d82c593 | | Chameleon | 39 | Not recorded | 6a0beeb9dbc836b87d82c593 | ### LAC Overall Accuracy: 90. Result ID: 6a0c28e4fbc373592c027016. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Unified Multimodal","Prompt Type":"LLM-rewritten"}. This record does not assert independent reproduction. ### Mogao Overall Accuracy: 89. Result ID: 6a0c28e4fbc373592c027012. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Unified Multimodal","Prompt Type":"LLM-rewritten"}. This record does not assert independent reproduction. ### BAGEL Overall Accuracy: 86. Result ID: 6a0c28e4fbc373592c027015. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Architecture":"Unified Multimodal","Prompt Type":"LLM-rewritten"}. This record does not assert independent reproduction. ### FLUX.1-dev Overall Accuracy: 82. Result ID: 6a0c28e4fbc373592c027009. [Reporting paper]() Recorded evaluation configuration: {"Architecture":"Generation Only","Prompt Type":"LLM-rewritten"}. This record does not assert independent reproduction. ### LAC Overall Accuracy: 82. Result ID: 6a0c28e4fbc373592c027014. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Unified Multimodal","Prompt Type":"Original"}. This record does not assert independent reproduction. ### MetaQuery-XL Overall Accuracy: 80. Result ID: 6a0c28e4fbc373592c027010. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Architecture":"Unified Multimodal","Prompt Type":"LLM-rewritten"}. This record does not assert independent reproduction. ### Janus-Pro-7B Overall Accuracy: 80. Result ID: 6a0c28e4fbc373592c027011. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Architecture":"Unified Multimodal","Prompt Type":"Original"}. This record does not assert independent reproduction. ### BAGEL Overall Accuracy: 77. Result ID: 6a0c28e4fbc373592c027013. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Architecture":"Unified Multimodal","Source":"Reproduced","Prompt Type":"Original"}. This record does not assert independent reproduction. ### Show-o2 Overall Accuracy: 76. Result ID: 6a0c28e4fbc373592c02700f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Unified Multimodal","Prompt Type":"LLM-rewritten"}. This record does not assert independent reproduction. ### SD3-Medium Overall Accuracy: 74. Result ID: 6a0c28e4fbc373592c027008. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Generation Only","Prompt Type":"Original"}. This record does not assert independent reproduction. ### DALL-E 3 Overall Accuracy: 67. Result ID: 6a0c28e4fbc373592c027007. [Reporting paper]() Recorded evaluation configuration: {"Architecture":"Generation Only","Prompt Type":"Original"}. This record does not assert independent reproduction. ### Emu3-Gen Overall Accuracy: 66. Result ID: 6a0c28e4fbc373592c02700e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Unified Multimodal","Prompt Type":"LLM-rewritten"}. This record does not assert independent reproduction. ### Transfusion Overall Accuracy: 63. Result ID: 6a0c28e4fbc373592c02700d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Unified Multimodal"}. This record does not assert independent reproduction. ### TokenFlow-XL Overall Accuracy: 55. Result ID: 6a0c28e4fbc373592c02700c. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Architecture":"Unified Multimodal","Prompt Type":"Original"}. This record does not assert independent reproduction. ### SEED-X Overall Accuracy: 49. Result ID: 6a0c28e4fbc373592c02700b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Unified Multimodal","Prompt Type":"Original"}. This record does not assert independent reproduction. ### PixArt-Σ Overall Accuracy: 48. Result ID: 6a0c28e4fbc373592c027006. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Architecture":"Generation Only","Prompt Type":"Original"}. This record does not assert independent reproduction. ### Chameleon Overall Accuracy: 39. Result ID: 6a0c28e4fbc373592c02700a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Architecture":"Unified Multimodal"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.