inclusionAI ORG

LLaDA2.0 is the diffusion language model series developed by InclusionAI team, Ant Group.

No description.

No description.

Ming - facilitating advanced multimodal understanding and generation capabilities built upon the Ling LLM.

[ICML 2026] ZwZ model family: SOTA fine-grained perception performace; ZoomBench: a new challenging perception benchmark

Search, understand, reproduce, and improve an idea with ease

UI-Venus is a native UI agent designed to perform precise GUI element grounding and effective navigation using only screenshots as input.

LLaDA2.0-Uni: Understanding and Generation the World.

An Open-Source Large-Scale Reinforcement Learning Project for Search Agents

[ICLR 2026] Taming large-scale few-step training with self-adversarial flows! 👏🏻

CUDA kernels for linear attention variants, written in CuTe DSL and CUTLASS C++.

dInfer: An Efficient Inference Framework for Diffusion Language Models

Ming-UniAudio: Speech LLM for Joint Understanding, Generation and Editing with Unified Representation

Standardized environment infrastructure for Agentic AI development.

Ling-V2 is a MoE LLM provided and open-sourced by InclusionAI.

Ming-omni-tts: Simple and Efficient Unified Generation of Speech, Music, and Sound with Precise Control

Easy and Efficient dLLM Fine-Tuning

Ling is a MoE LLM provided and open-sourced by InclusionAI.

Distributed agent coordination platform where agents live, connect, coordinate, execute, and evolve together.

A high-performance RL training-inference weight synchronization framework, designed to enable second-level parameter updates from training to inference in RL workflows

No description.

Code release for Ming-UniVision: Joint Image Understanding and Geneation with a Continuous Unified Tokenizer

A unified suite for generating elite reasoning problems and training high-performance LLMs, including pioneering attention-free architectures

Agentic Learning Powered by AWorld

A NCCL extension library, designed to efficiently offload GPU memory allocated by the NCCL communication library.

Ring is a reasoning MoE LLM provided and open-sourced by InclusionAI, derived from Ling.

Ring-V2 is a reasoning MoE LLM provided and open-sourced by InclusionAI.

No description.

No description.

No description.

A high-performance kernel library for LLM training

[ICLR 2026] AudioMCQ: A 571k audio multiple-choice question dataset for post-training Large Audio Language Models with dual CoT annotations and audio-contribution filtering. 🏆 1st place in DCASE 2025 Challenge.

No description.

HeartBench is an evaluation benchmark for the psychological and social sciences field, designed to transcend traditional knowledge and reasoning assessments. It focuses on measuring large language models' (LLMs) anthropomorphic capabilities in human-computer interactions, covering dimensions such as personality, emotion, social skills, and ethics.

M2-Reasoning: Empowering MLLMs with Unified General and Spatial Reasoning

An easy-to-use, fast toolkit to scale up RL post-training on a single node.

No description.

Mixture-of-Basis-Experts for Compressing MoE-based LLMs

Programmable datacenter-scale infrastructure for Agents.

ABench is an evolving open-source benchmark suite designed to rigorously evaluate and enhance Large Language Models (LLMs) on complex cross-domain tasks.

Official repo for "TC-AE: Unlocking Token Capacity for Deep Compression Autoencoders"

No description.

PIBench is an open-source benchmark for evaluating AI coding agents on realistic, end-to-end payment integration tasks.

No description.

No description.

No description.

No description.

[NeurIPS 2025] ARGenSeg: Image Segmentation with Autoregressive Image Generation Model

Universal Manipulation Exoskeleton: Learning Compliant Whole-body Policies with Real-time Torque Feedback

🔥 inclusionAI Official Website and Blog.

A sandbox runtime daemon.

Read-only FUSE filesystem for raw disk images and Nydus RAFS images.

No description.

SGLang is a high-performance serving framework for large language models and multimodal models.

Gorilla: Training and Evaluating LLMs for Function Calls (Tool Calls)

A high-throughput and memory-efficient inference and serving engine for LLMs

No description.