mllm-reasoning 开源项目推荐（18 个）

yaotingwangofficial /Awesome-MCoT

Multimodal Chain-of-Thought Reasoning: A Comprehensive Survey

NVlabs /SpatialClaw

[NeurIPS 2026] SpatialClaw: Rethinking Action Interface for Agentic Spatial Reasoning

ritzz-ai /GUI-R1

GUI-R1的官方实现：一种通用的R1风格视觉语言动作模型，适用于图形界面代理

THUSI-Lab /Hstar

[CVPR 2026] Thinking in 360°: Humanoid Visual Search in the Wild

JavisVerse /Awesome-AVI

Awesome Audio-Visual Intelligence, Survey of Audio-Visual Intelligence

Wild-Cooperation-Hub /Awesome-MLLM-Reasoning-Benchmarks

A Comprehensive Survey on Evaluating Reasoning Capabilities in Multimodal Large Language Models.

luo-junyu /FinMME

[ACL 2025] FinMME: Benchmark Dataset for Financial Multi-Modal Reasoning Evaluation

CharlesPikachu /FreeGPTHub

FreeGPTHub: A truly free, unified GPT API gateway—unlike “fake-free” alternatives that hide paywalls behind quotas, trials, or mandatory top-ups. (真正免费的GPT统一接口，支持讯飞星火知识大模型，百度文心大模型，OpenAI ChatGPT，智谱大模型，幻方量化深度求索，字节豆包，阿里通义千问，稀宇科技MiniMax，魔搭社区开源模型等)

Kun-Xiang /AtomThink

[TPAMI 2026] Offical Repository of "AtomThink: Multimodal Slow Thinking with Atomic Step Reasoning"

YutingLi0606 /Vision-Matters

(ArXiv25) Vision Matters: Simple Visual Perturbations Can Boost Multimodal Math Reasoning

falonss703 /Awesome-Uncertainty-based-Reinforcement-Learning

🔥🔥🔥Latest Papers, Codes on Uncertainty-based RL

manglu097 /Chiron-o1

[NIPS 2025] Chiron-o1: Igniting Multimodal Large Language Models towards Generalizable Medical Reasoning via Mentor-Intern Collaborative Search

Jorffy /NoteMR

[CVPR 2025] Code for "Notes-guided MLLM Reasoning: Enhancing MLLM with Knowledge and Visual Notes for Visual Question Answering".

SkyworkAI /CSVQA

A Multimodal Benchmark for Evaluating Scientific Reasoning Capabilities of VLMs

stvngo /GamiBench

Multi-modal and Vision Language Model Spatial Reasoning Benchmark

vulab-AI /YESBUT-v2

We introduce the YesBut-v2, a benchmark for assessing AI's ability to interpret juxtaposed comic panels with contradictory narratives. Unlike existing benchmarks, it emphasizes visual understanding, comparative reasoning, and social knowledge.

go34loj /bravo_benchmark

BRAVO (Building Regulation Answering & Visual Observation) Bench is the first benchmark designed to evaluate the capability of MLLMs to perform compliance checking based on BIM-derived scenes.

XueTeng0221 /EARS

an MLLM-based trust voice reasoning and retrieval system