[claude] Add vllm-mlx as high-performance local inference backend (#1069) (#1089)

Co-authored-by: Claude (Opus 4.6) <claude@hermes.local>
Co-committed-by: Claude (Opus 4.6) <claude@hermes.local>

src/infrastructure/router/cascade.py

@@ -310,6 +310,22 @@ class CascadeRouter:
                 logger.debug("Ollama provider check error: %s", exc)
                 return False
 
+        elif provider.type == "vllm_mlx":
+            # Check if local vllm-mlx server is running (OpenAI-compatible)
+            if requests is None:
+                return True
+            try:
+                base_url = provider.base_url or provider.url or "http://localhost:8000"
+                # Strip /v1 suffix — health endpoint is at the root
+                server_root = base_url.rstrip("/")
+                if server_root.endswith("/v1"):
+                    server_root = server_root[:-3]
+                response = requests.get(f"{server_root}/health", timeout=5)
+                return response.status_code == 200
+            except Exception as exc:
+                logger.debug("vllm-mlx provider check error: %s", exc)
+                return False
+
         elif provider.type in ("openai", "anthropic", "grok"):
             # Check if API key is set
             return provider.api_key is not None and provider.api_key != ""
@@ -619,6 +635,14 @@ class CascadeRouter:
                 temperature=temperature,
                 max_tokens=max_tokens,
             )
+        elif provider.type == "vllm_mlx":
+            result = await self._call_vllm_mlx(
+                provider=provider,
+                messages=messages,
+                model=model or provider.get_default_model(),
+                temperature=temperature,
+                max_tokens=max_tokens,
+            )
         else:
             raise ValueError(f"Unknown provider type: {provider.type}")
 
@@ -815,6 +839,48 @@ class CascadeRouter:
             "model": response.model,
         }
 
+    async def _call_vllm_mlx(
+        self,
+        provider: Provider,
+        messages: list[dict],
+        model: str,
+        temperature: float,
+        max_tokens: int | None,
+    ) -> dict:
+        """Call vllm-mlx via its OpenAI-compatible API.
+
+        vllm-mlx exposes the same /v1/chat/completions endpoint as OpenAI,
+        so we reuse the OpenAI client pointed at the local server.
+        No API key is required for local deployments.
+        """
+        import openai
+
+        base_url = provider.base_url or provider.url or "http://localhost:8000"
+        # Ensure the base_url ends with /v1 as expected by the OpenAI client
+        if not base_url.rstrip("/").endswith("/v1"):
+            base_url = base_url.rstrip("/") + "/v1"
+
+        client = openai.AsyncOpenAI(
+            api_key=provider.api_key or "no-key-required",
+            base_url=base_url,
+            timeout=self.config.timeout_seconds,
+        )
+
+        kwargs: dict = {
+            "model": model,
+            "messages": messages,
+            "temperature": temperature,
+        }
+        if max_tokens:
+            kwargs["max_tokens"] = max_tokens
+
+        response = await client.chat.completions.create(**kwargs)
+
+        return {
+            "content": response.choices[0].message.content,
+            "model": response.model,
+        }
+
     def _record_success(self, provider: Provider, latency_ms: float) -> None:
         """Record a successful request."""
         provider.metrics.total_requests += 1