ASI-Evolve
ASI-Evolve is an autonomous AI research agent framework that iteratively executes a knowledge-hypothesis-experiment-analysis loop to discover novel solutions across domains. Designed initially for AI research tasks like neural architecture design and algorithm optimization, it employs three coordinating agents (Researcher, Engineer, Analyzer) and two memory systems (Cognition Store and Experiment Database) to autonomously propose, test, and learn from hundreds of candidates. The framework has validated its ability to achieve state-of-the-art results in neural architecture search, pretraining data curation, reinforcement learning algorithms, and biomedical modeling. Developers can adapt it to any domain where programmatic solutions can be evaluated automatically, from infrastructure optimization to scientific discovery.
Key Features
Use Cases
- 01Neural architecture search and linear attention mechanism design for transformer models
- 02Pretraining data curation pipelines to improve language model performance
- 03Reinforcement learning algorithm discovery with novel optimization mechanisms
- 04Drug-target interaction model optimization for biomedical cold-start scenarios
- 05LLM serving scheduler optimization to maximize throughput while maintaining latency SLAs
- 06Manufacturing quality control classifier tuning and defect detection pipeline evolution
Related Agents
View morehermes-agent
The agent that grows with you
agency-agents
A complete AI agency at your fingertips - From frontend wizards to Reddit community ninjas, from whimsy injectors to reality checkers. Each agent is a specialized expert with personality, processes, and proven deliverables.
openinterpreter
A coding agent for open models like Kimi K3
cline
Autonomous coding agent as an SDK, IDE extension, or CLI assistant.
ASI-Evolve — FAQ
What is ASI-Evolve?+
ASI-Evolve is an autonomous AI agent framework that closes the loop between knowledge retrieval, hypothesis generation, experimentation, and analysis. It iteratively searches solution spaces to discover optimized programs, algorithms, or pipelines across any domain where automated evaluation is possible.
How do I install ASI-Evolve?+
Clone the repository from GitHub, install Python 3.10+ dependencies via pip install -r requirements.txt, and ensure bash and python3 are in your system path. You'll also need access to an OpenAI-compatible API endpoint (GPT-4, Claude, or local models).
What are the prerequisites for using ASI-Evolve?+
You need a problem where better code yields better outcomes, an evaluation script that scores candidate programs, and some domain knowledge to seed the cognition store. An API key for an LLM provider (OpenAI, Anthropic, Google, or a local LiteLLM endpoint) is required.
Is ASI-Evolve free to use?+
The framework itself is open-source and free. However, running it incurs API costs for the underlying language model calls (GPT-4, Claude, etc.), which can vary depending on the number of evolution steps and parallel workers configured.
Which AI clients or platforms does ASI-Evolve work with?+
ASI-Evolve is a standalone framework that runs from the command line and supports any OpenAI-compatible API endpoint, including GPT-4, Claude via LiteLLM, Gemini, or locally hosted models. It is not a plugin for Claude Desktop, Cursor, or similar assistant clients.
How many evolution rounds do I need to run?+
The number of steps depends on problem complexity. The included circle-packing demo reaches SOTA-level results in ~17 rounds, while production tasks like neural architecture search or scheduler optimization typically use 40–200 steps for comprehensive exploration.
How do I install ASI-Evolve?+
Open the source repository on GitHub and follow its README. ASI-Evolve is a agent — MCP Agents Market links you directly to the official repo.
Is ASI-Evolve free?+
ASI-Evolve is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.