Introduction

Overview

The AI-innovator suite aims to endow the AI agents with the ability to conduct innovative LLM research.

The AI-innovator suite includes:

• A benchmark of 20 LLM research tasks
• A ResearchGym execution environment
• A interactive human-agent interface in real-time
• A model registry for storing and sharing models

Model Support

OpenAI (through Azure)

• GPT-5

Anthropic (through OpenRouter)

• Claude Sonnet 4

Zhipu AI

• GLM-4.5

Kimi

• Kimi-K2

Want to contribute?

Visit our InnovatorBench’s GitHub repository to contribute to the project.