Stop paying $20/mo for Claude or Copilot. local-coder checks what your machine can run, then generates a benchmark-backed Ollama + Continue.dev config pack — working in under an hour.
Profiles your RAM and GPU VRAM and picks the model that's the sweet spot for your box — not a generic blog default that's too big or too small.
Generates a wired-up Continue.dev config for VS Code, Cursor, and JetBrains, plus an Ollama pull script and ready-to-use prompt templates.
Codebase-aware retrieval is configured out of the box, so the assistant answers with context from your actual project.
Benchmark-backed model picks, not guesswork.
| Tier | Budget | Primary model |
|---|---|---|
| Tiny | < 6GB | qwen2.5-coder:3b |
| Small | 8–16GB | qwen2.5-coder:7b |
| Medium | 16–32GB RAM / 12–16GB VRAM | qwen2.5-coder:14b |
| Large | 32GB+ / 24GB+ VRAM | qwen2.5-coder:32b + deepseek-coder-v2 |
Free forever to try. One-time Pro — no subscription required.