@orqenix/llm-adapter-ollama

Local LLM adapter using the Ollama HTTP API. Default model: qwen2.5:7b.

Why this is the OSS default

CR v7.1 chooses Qwen 2.5 7B via Ollama as the OSS default because:

1. Runs fully offline on consumer hardware (8GB+ VRAM or CPU)
2. Apache-2.0 weights, no API key, no telemetry
3. Strong instruction following at small size for the prompt-rewriter use case

Quick start

import { OllamaAdapter } from "@orqenix/llm-adapter-ollama";

const llm = new OllamaAdapter({ model: "qwen2.5:7b" });
const r = await llm.complete({ messages: [{ role: "user", content: "hi" }] });
console.log(r.content);

Charter gate: G8 LLM Adapter Contract.