---
id: "nr-hugging-face-space-runs-minicpm5-2b-coding-agent-in-0f5bd4"
slug: "hugging-face-space-runs-minicpm5-2b-coding-agent-in-0f5bd4"
title: "Hugging Face Space runs MiniCPM5-2B coding agent in the browser"
observed_at: "2026-09-18T13:40:45.45798+00:00"
record_date: "2026-09-18"
date_kind: "observed_at"
novelty: "new"
source_platform: "huggingface.co"
topics: ["local-inference","coding-agents"]
entities: ["MiniCPM5-2B","Transformers.js","Hugging Face"]
related_patterns: []
source_url: "https://huggingface.co/spaces/victor/MiniCPM5-2B-WebGPU-Pi"
source_urls: ["https://huggingface.co/spaces/victor/MiniCPM5-2B-WebGPU-Pi"]
basket: {"id":"0325e6b8-f527-4f9b-bfa3-3f30e57f380b","revision":1,"cluster_id":"0f5bd44e-0c67-4f11-ba59-d8a02f3e5cc6"}
type: "raw_signal"
description: "A Hugging Face Space runs a Pi-based coding agent on OpenBMB's MiniCPM5-2B entirely in the browser using WebGPU inference and an in-browser Bash shell."
why_it_matters: "For engineers, it demonstrates a fully local agent loop in which prompts and files stay on the device. The page does not report task success rates, speed or which GPUs can run it."
verification_level: "source-inspected"
signal_type: "newsroom_basket_signal"
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "signal:hugging-face-space-runs-minicpm5-2b-coding-agent-in-0f5bd4"
retrieval_nugget: "A Hugging Face Space runs a Pi-based coding agent on OpenBMB's MiniCPM5-2B entirely in the browser using WebGPU inference and an in-browser Bash shell. A Hugging Face Space by the user victor, MiniCPM5-2B WebGPU Pi, runs a coding agent entirely in the browser. It uses OpenBMB's MiniCPM5-2B model with inference on the local GPU and a Bash shell in the"
status: "published"
visuals: []
routes: {"html":"https://newruntime.com/signals/hugging-face-space-runs-minicpm5-2b-coding-agent-in-0f5bd4/","markdown":"https://newruntime.com/signals/hugging-face-space-runs-minicpm5-2b-coding-agent-in-0f5bd4.md","json":"https://newruntime.com/signals/hugging-face-space-runs-minicpm5-2b-coding-agent-in-0f5bd4.json"}
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
---

# Hugging Face Space runs MiniCPM5-2B coding agent in the browser

## Retrieval answer

A Hugging Face Space runs a Pi-based coding agent on OpenBMB's MiniCPM5-2B entirely in the browser using WebGPU inference and an in-browser Bash shell. A Hugging Face Space by the user victor, MiniCPM5-2B WebGPU Pi, runs a coding agent entirely in the browser. It uses OpenBMB's MiniCPM5-2B model with inference on the local GPU and a Bash shell in the

A Hugging Face Space by the user victor, MiniCPM5-2B WebGPU Pi, runs a coding agent entirely in the browser. It uses OpenBMB's MiniCPM5-2B model with inference on the local GPU and a Bash shell in the browser, downloading about 1.84 GB of weights once and caching them. The stack lists Transformers.js, ONNX Runtime, the Pi agent toolkit and just-bash.

For engineers, it demonstrates a fully local agent loop in which prompts and files stay on the device. The page does not report task success rates, speed or which GPUs can run it.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Local Inference](https://newruntime.com/topics/local-inference.md): Explore the local-inference topic hub.
- topic hub: [Coding Agents](https://newruntime.com/topics/coding-agents.md): Explore the coding-agents topic hub.
- collection: [Raw Signals](https://newruntime.com/signals.json): Open every source-linked raw signal.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)