Add model selection and local RAG

benchmark_models.py

@@ -0,0 +1,35 @@
+from __future__ import annotations
+
+import time
+
+import ollama
+
+
+MODELS = ["qwen2.5:7b", "llama3.1:8b", "qwen2.5:14b"]
+PROMPT = (
+    "Answer in Korean in exactly three bullets. Compare your usefulness for "
+    "local coding assistance, Korean Q&A, and careful reasoning."
+)
+
+
+def main() -> None:
+    for model in MODELS:
+        started = time.perf_counter()
+        try:
+            response = ollama.chat(
+                model=model,
+                messages=[
+                    {"role": "system", "content": "Be concise and practical."},
+                    {"role": "user", "content": PROMPT},
+                ],
+            )
+            elapsed = time.perf_counter() - started
+            text = response["message"]["content"].strip().replace("\n", " ")
+            print(f"{model}\t{elapsed:.1f}s\t{text[:180]}")
+        except Exception as exc:
+            elapsed = time.perf_counter() - started
+            print(f"{model}\t{elapsed:.1f}s\tERROR: {exc}")
+
+
+if __name__ == "__main__":
+    main()