Research map / Image generation / Text-to-image diffusion
FLUX-Reason-6M & PRISM-Bench: A Million-Scale Text-to-Image Reasoning Dataset and Comprehensive Benchmark
Rongyao Fang, Aldrich Yu, Chengqi Duan, Linjiang Huang, Shuai Bai, Yuxuan Cai, Kun Wang, Si Liu, Xihui Liu, Hongsheng Li
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| scale model generation from SD1.5 to SDXL | Stable Diffusion series · PRISM-Bench overall (GPT-4.1) | average score: 44.2 -> 60.4 | Improved |
| use closed-source models instead of best open-source model | GPT-Image-1 vs Qwen-Image · PRISM-Bench overall (GPT-4.1) | average score: 79.9 -> 86.3 | Improved |
| autoregressive architecture (Bagel, JanusPro) vs diffusion on text rendering | Bagel / JanusPro-7B · PRISM-Bench Text rendering | average score: 37.4 / 26.1 vs FLUX.1-dev 56.3 | Hurt |
| add chain-of-thought to Bagel (Bagel-CoT) | Bagel · PRISM-Bench overall (GPT-4.1) | average score: 65.1 -> 65.2 | No clear effect |
| evaluate with Qwen2.5-VL-72B instead of GPT-4.1 as judge | PRISM-Bench · model ranking | overall score: GPT-Image-1 86.3 -> 80.7 | No clear effect |
| use Chinese prompts (PRISM-Bench-ZH) instead of English | Bagel · PRISM-Bench overall (GPT-4.1) | average score: 65.1 -> 65.4 | No clear effect |
Closest papers by meaning
- OneIG-Bench: Omni-dimensional Nuanced Evaluation for Image Generation (NeurIPS 2025 · similarity 0.89)
- MMMG: A Massive, Multidisciplinary, Multi-Tier Generation Benchmark for Text-to-Image Reasoning (NeurIPS 2025 · similarity 0.88)
- GIR-Bench: Versatile Benchmark for Generating Images with Reasoning (ICLR 2026 · similarity 0.88)
- MMTIT-Bench: A Multilingual and Multi-Scenario Benchmark with Cognition-Perception-Reasoning Guided Text-Image Machine Translation (CVPR 2026 · similarity 0.87)
- When Visualizing is the First Step to Reasoning: MIRA, a Benchmark for Visual Chain-of-Thought (CVPR 2026 · similarity 0.87)
- ARC Is a Vision Problem! (CVPR 2026 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
