AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

OctoT2I: A Self-Evolving Agentic Text-to-Image Router

Xu Jiang, Bin Chen, Gehui Li, Yule Duan, Ronggang Wang, Jian Zhang

CVPR 2026 · Image generation › Text-to-image diffusion · approach: T2I evaluation and controllable adapters · 3 citations (OpenAlex)

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 changedWhereMeasuredEffect
add self-evolved knowledge instead of GPT internal knowledge (ablation)OctoT2I · GenEvaloverall score: 0.85 -> 0.96Improved
add self-evolved knowledge instead of hand-crafted prior (ablation)OctoT2I · GenEvaloverall score: 0.93 -> 0.96Improved
add knowledge-driven decision policy instead of random tool selection (ablation)OctoT2I · T2I-CompBench++average score: 0.5379 -> 0.6618Improved
add exploration space pruning (ablation)OctoT2I · GenEvalexplored prompts: 1270 -> 370Improved
add exploration space pruning (ablation)OctoT2I · GenEvalexploration time (s): 6857.4 -> 2328.7Improved
add exploration space pruning (ablation)OctoT2I · GenEvaloverall score: 0.96 -> 0.96No clear effect
raise quality threshold theta (ablation)OctoT2I · GenEvalperformance and latency: most significant gain up to theta=0.8Mixed
add one new tool (FLUX.1-dev) with 60 exploration promptsOctoT2I · WISEaverage score: 0.54 -> 0.61Improved
add second new tool (gpt-image-1)OctoT2I · WISEaverage score: 0.61 -> 0.71Improved
swap controller LLM to GPT-5OctoT2I · WISEaverage score: 0.54 -> 0.60Improved
swap controller LLM to Qwen3-VL-235B-A22BOctoT2I · WISEaverage score: 0.54 -> 0.55No clear effect
use OctoT2I instead of Flow-GRPOOctoT2I · GenEval (efficiency)average time (s): 19.07 -> 10.02Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy