AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

TIR-Bench: A Comprehensive Benchmark for Agentic Thinking-with-Images Reasoning

Ming Li, Jike Zhong, Shitian Zhao, Haoquan Zhang, Shaoheng Lin, Yuxiang Lai, Chen Wei, Konstantinos Psounis, Kaipeng Zhang

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Chain-of-thought multimodal reasoning · 34 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
enable code interpreter (tool use) (ablation)o3 · TIR-Bench all tasksaccuracy: 26.9 -> 46.0Improved
add PyVision thinking-with-images tool frameworkGPT-4.1 · TIR-Bench all tasksaccuracy: 18.8 -> 31.8Improved
enable code interpreter (ablation)o3 · TIR-Bench Word Searchaccuracy: 4.0 -> 64.0Improved
enable code interpreter (ablation)o3 · TIR-Bench Rotation Gameaccuracy: 33.3 -> 77.3Improved
enable code interpreter (ablation)o3 · TIR-Bench Rotated OCRaccuracy: 8.3 -> 53.3Improved
enable code interpreter (ablation)o3 · TIR-Bench Proportion VQAaccuracy: 34.2 -> 31.7Hurt
enable code interpreter (ablation)o3 · TIR-Bench Jigsaw Gameaccuracy: 10.8 -> 16.4No clear effect
scale open-source model size 3B to 72BQwen2.5-VL · TIR-Bench all tasksaccuracy: 17.7 -> 19.7No clear effect
scale InternVL3 from 8B to 78BInternVL3 · TIR-Bench all tasksaccuracy: 16.9 -> 21.4Improved
use explicit prompt to test each degree choice (prompt 3) (ablation)most models · Rotation Game function callingaccuracyImproved
tool-use SFT instead of direct SFT (ablation)Qwen2.5-VL-7B · Rotated OCRaccuracyImproved
scale direct SFT data from 1k to 15k (ablation)Qwen2.5-VL-7B direct SFT · Rotated OCRaccuracyNo clear effect

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