Local Model
An LLM you run on your own hardware via Ollama, LM Studio, or llama.cpp instead of the cloud.
- Keeps prompts and code on your machine; offline-capable.
- No per-token cost; needs 16-32 GB RAM minimum.
- Weaker reasoning than frontier API models.
When to use: For sensitive codebases, air-gapped envs, or token-cost-dominated workloads.
Example: `ollama run qwen2.5-coder:7b` downloads and runs a 7B coding model; point Continue at `localhost:11434`.