JoinGrove

Offline AI Providers

Source: product.json → joingroveOfflineProviders
Generated: 01/09/2026

Key Concepts

  • Default provider: Ollama at http://localhost:11434.
  • Tiered models: Lite tier for ≤8 GB RAM; standard for 16 GB+ when adaptive performance is enabled.
  • Zero subscription: Local models work without GitHub Copilot cloud subscription via BYOK.

API / Reference

Recommended models

Model RAM tier Notes
qwen2.5-coder:1.5b ≤ 8 GB Lite / low-spec laptops
qwen2.5-coder:7b ≤ 16 GB Standard default

All recommended pulls

Model
qwen2.5-coder:1.5b
qwen2.5-coder:7b
llama3.2
deepseek-r1:7b

Settings

Setting Default
Ollama URL http://localhost:11434
Default model qwen2.5-coder:7b
Lite model qwen2.5-coder:1.5b
Config file ~/.joingrove/User/chatLanguageModels.json

Setup & Installation

Reference tables are generated from canonical sources (product.json, command registrations, egress policy). After changing those sources, regenerate:

node scripts/generate-docs-reference.mjs

Next: Run npm run verify-docs to confirm tables match sources.

Code Examples

npm run generate-docs-reference

Next: Open the docs hub with joingrove.openDocsHub or browse topics with joingrove.openDocsTopic.

How It Connects