FreeLLMOffline coding agent runs on laptop with OpenCode, Ollama and Qwen3.5 9B
A principal engineer demonstrates a fully offline workflow using OpenCode, Ollama and the Qwen3.5 9B model on a Mac laptop.
Correction, 30 October 2026: an unrelated GitHub project was wrongly listed as a source here, and we removed it. The article text is unchanged.
Rittika Jindal, a Principal Engineer at Thomson Reuters, set up a local coding assistant that works without an internet connection.
The workflow combines OpenCode 1.19.4, which reads project files and proposes edits, with Ollama running the open‑weight Qwen3.5 9B model on the same machine.
Qwen3.5 9B weighs about 6.6–7.6 GB to download and requires 12–16 GB of RAM, fitting on a typical laptop with enough free memory for a 64k‑token context.
Published scores show Qwen3.5 9B achieving 65.6 on LiveCodeBench v6 and 66.1 on BFCL‑V4, outperforming the 4‑B variant used in the same article.
Benchmarks from a cloud‑based study reported OpenCode achieving a 50 % pass rate on Terminal‑Bench Pro tasks, though token usage was higher than competing agents.
Ce que cela signifie pour vous
Running a coding agent fully offline lets developers work in remote locations with unreliable connectivity, preserving productivity while keeping code and data local. The availability of open‑weight models like Qwen3.5 9B demonstrates that powerful assistance can be achieved without relying on paid cloud APIs, supporting the push for free and self‑hosted AI tools.
Sources
- Coding From the Himalayas With an Offline AI Agent, Latent Space, 27 Oct 2026
Rédigé automatiquement à partir des sources ci‑dessus, puis vérifié par notre code : chaque nombre et chaque nom doivent y figurer, et aucune phrase ne doit copier une source. Une erreur ? Signalez‑la.
Quelle IA pouvez‑vous utiliser gratuitement aujourd’hui ?
Tous les modèles IA gratuits, classés par votes humains anonymes, avec la mémoire et la vitesse dont ils ont besoin sur votre propre GPU.