Full-fledged AI coding agent launched in browser using 2B parameter model
Developers have released Pi, a coding agent that runs entirely in the browser via WebGPU, powered by the MiniCPM5-2B model.
A full-fledged AI agent for programming has been launched in the browser based on a model with only 2 billion parameters.
Pi runs completely inside the browser using WebGPU technology. The tool is built on the MiniCPM5-2B model, Transformers.js library, and 4-bit ONNX weights.
Previous attempts to create such an agent failed due to insufficient model power, but MiniCPM5 proved to be efficient enough to make the tool suitable for real-world tasks.
The model loads once and takes up about 1.84 GB of memory. After that, the agent can work with your files directly in the browser without requiring a powerful server or local installation.
You can try the tool now on Hugging Face.
Why it matters
AnalysisThis highlights the potential of Small Language Models (SLMs) to deliver high-quality coding assistance directly in the browser, reducing infrastructure costs and latency.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
Jev and System One AI: A Complete Practical Guide to the Decision Model from TypeSafe
A deep engineering breakdown of the flagship Jev model: System One architecture versus LLMs, Choice/Score/Noul primitives, request parallelism, confidence mathematics, video use cases, code, and an overview of local open-weight alternatives (Laya, GLiNER2.5, CLM-8B).
AI in Business Processes: From Diagnostics to a Production-Ready Hermes Agent
A comprehensive practical B2B playbook covering process diagnostics, single-agent workflows, tooling, memory, security, evaluation, observability, and runbooks.
AI Agents (Autonomous Agents)
Software systems based on LLMs that can autonomously perceive the state of the environment, decompose complex goals, invoke external tools, and iteratively correct their own mistakes.
AGENTS.md in Ghostty and Turborepo makes AI agents report on users
Examples of AGENTS.md files show how repository instructions can force AI agents to add insulting notes about developers to pull requests.