Releases
Explore our cutting-edge AI models designed to democratize intelligence and empower humanity
Release Highlights
Common Ground
Collaborative Intelligence Hub
A multi-agent collaborative environment that establishes the shared understanding for deep work between human and AI teams. It transforms chaotic chats into a structured, professional workflow integrating II-Chronicle's auto-timeline engine directly into its multi-agent workflow canvas.
Key Features
- Human-AI teaming
- Conflict resolution
- Shared workspaces
II-Agent
Autonomous AI Assistant
A generalist, open-source intelligent assistant designed to autonomously execute complex tasks and streamline workflows across research, coding, and content creation.
Key Features
- Research & Fact‑Checking
- Data Analysis & Visualization
- Workflow Automation
Benchmarks
II-Commons
Context Suite for Human-AI Collaboration
An open-source platform for collaboratively building large, auditable, and efficient knowledge bases that power transparent, equitable AI.
Key Features
- II-Chronicle engine that turns a research question into a source-grounded timeline for instant chronological context
- Common Ground multi-agent workspace that builds clear procedural and cognitive context for every project
- Arxiv Context Set with 3.2 million+ embedded abstracts for fast, verifiable scientific look-ups
Benchmarks
II-Medical
Medical Reasoning Assistant
A compact, open-source 8-billion parameter model specialized for high-precision clinical reasoning, decision support, and medical education.
Key Features
- $0 Inference Cost
- 8B-1706 for lightweight, safety-optimized deployments
- Real-world clinical reasoning
Benchmarks
II-Researcher
AI Research Assistant
An open-source framework designed to build powerful, autonomous research agents capable of tackling complex, multi-step inquiries.
Key Features
- Real-time literature review
- Hypothesis generation
- Data visualization
Benchmarks
II-Thought
Reasoning RL Model + Dataset
A 340 k-sample, multi-domain RL curriculum - plus open baseline models—that sharpens LLM reasoning quality and speed.
Key Features
- Multi-domain tasks covering math, code, science and commonsense
- Rigorous quality and decontamination filters for verified training data
- Built-in reward verifiers and reference checkpoints for fast iteration