Running a fully local AI coding stack: Pi agent plus 58GB Qwen coder on an M3 Max
julsimon · x · 2026-11-11
julsimon demos a cloud-free, fully local AI coding setup:
- Agent: the open-source coding agent Pi, used first in plan mode, then in build mode
- Model: Qwen3.8 Flash Next Coder, compressed to 58 GB by ISTA-DASLab and hosted on Hugging Face as GGUF
- Hardware: Apple M3 Max with 48 GB of RAM
The workflow: write the spec in plain English, then develop with TDD. It's a concrete proof that consumer laptops can now run capable open-source coding agents end to end.
Related event: Developers Run Pi Coding Agent Fully Locally with Qwen Models(4 posts)→
More from coding & agent
- Open-source Agent Standup taps Claude's new Whiteboards and Sheets for agent standups and timesheets — stickygoose · 2026-11-11
- Nod details its agent evals: 98% task completion across 678 runs, zero of 216 memory injections succeed — rohanpaul_ai · 2026-11-11
- 9-year Rust veteran argues Zig beats Rust in the AI coding era — nicolascraske · 2026-11-11
- Cheap builder + frontier reviewer loop matches flagship on 66 tasks at ~1/16 the cost ($0.05 vs $0.72) — KrystalUnity · 2026-11-11
- As Hand-Coding Fades, the Edge Shifts to Research Insights LLMs Can't Supply — janusch_patas · 2026-11-11
- An LLM Hallucination Delivered 72 Eggs; Lesson: Always Review the Plan First — JFPuget · 2026-11-11