Make DEFAULT_MODEL env-var configurable (CAIC_DEFAULT_MODEL)

.env.example

@@ -29,8 +29,10 @@ CAIC_EMBED_MODEL=all-minilm:latest
 CAIC_AMQP_URL=amqp://caic:${RABBITMQ_PASSWORD}@rabbitmq:5672/caic
 
 # ── llama-server settings ────────────────────────────────────
-# Set this to your .gguf filename (must exist in ./models/)
+# GGUF filename (must exist in ./models/)
 LLAMA_MODEL=
+# Model name the app presents to clients (must match a loaded model)
+CAIC_DEFAULT_MODEL=qwen2.5-7b-instruct
 LLAMA_CTX_SIZE=4096
 LLAMA_N_GPU_LAYERS=0
 LLAMA_RPC_ENDPOINTS=

config.py

@@ -14,7 +14,7 @@ VERSION = "v0.22.1"
 OLLAMA_BASE = os.environ.get("OLLAMA_BASE", "http://localhost:11434")
 LLAMA_SERVER_BASE = os.environ.get("LLAMA_SERVER_BASE", "http://localhost:8081")
 SEARXNG_BASE = os.environ.get("CAIC_SEARXNG_BASE", "http://localhost:8888")
-DEFAULT_MODEL = "qwen2.5-7b-instruct"
+DEFAULT_MODEL = os.environ.get("CAIC_DEFAULT_MODEL", "qwen2.5-7b-instruct")
 COMPLETIONS_API_KEY = os.environ.get("CAIC_COMPLETIONS_API_KEY", "caic-sk-" + os.urandom(24).hex())
 MODEL_CONTEXT_LENGTH = 4096