Qwen3.8-27B enables fully local Computer Use via Codex

alexcovo_eth · x · 2026-08-16

Codex's native Computer Use feature now runs 100% locally on Qwen3.8-27B. This setup eliminates GPT subscription caps and cloud API bills. Users can pull the model via Ollama or llama.cpp, select it through OpenCodex, and execute screen actions, clicks, and workflows using local weights with zero cloud dependency.

Original post →

More from coding & agent

coding & agent channel →