feat: add embeddings endpoint and config for OpenAI-compatible text embeddings via gateway

Add POST /openai/v1/embeddings route in openai_gateway router, new config fields for embedding upstream URL/model/key/enabled toggle with defaults pointing to OpenRouter Qwen3 8B, INTERNAL_EMBED_MODEL constant in config.py, documentation in docs_api.py and README.md describing the molodetz~embed model mapping, and embed-call tracking in gateway metrics alongside existing chat/vision counters.

devplacepy/services/openai_gateway/service.py

@@ -130,6 +130,38 @@ class GatewayService(BaseService):
             help="Image-description LRU cache entries (0 disables caching).",
             group="Vision",
         ),
+        ConfigField(
+            "gateway_embed_enabled",
+            "Embeddings",
+            type="bool",
+            default=True,
+            help="Expose the embeddings model at /openai/v1/embeddings.",
+            group="Embeddings",
+        ),
+        ConfigField(
+            "gateway_embed_url",
+            "Embeddings URL",
+            type="url",
+            default=config.EMBED_URL_DEFAULT,
+            help="OpenAI-compatible embeddings endpoint requests are forwarded to.",
+            group="Embeddings",
+        ),
+        ConfigField(
+            "gateway_embed_model",
+            "Embeddings model",
+            type="str",
+            default=config.EMBED_MODEL_DEFAULT,
+            help="Embedding model sent upstream. Clients request it as molodetz~embed.",
+            group="Embeddings",
+        ),
+        ConfigField(
+            "gateway_embed_key",
+            "Embeddings API key",
+            type="str",
+            default="",
+            help="The key currently in use; falls back to the vision/OPENROUTER key on boot (embeddings default to OpenRouter). Editable.",
+            group="Embeddings",
+        ),
         ConfigField(
             "gateway_require_auth",
             "Require authentication",
@@ -216,6 +248,15 @@ class GatewayService(BaseService):
             minimum=0,
             group="Pricing",
         ),
+        ConfigField(
+            "gateway_embed_price_input_per_m",
+            "Embeddings price input / 1M ($)",
+            type="float",
+            default=config.EMBED_PRICE_INPUT_PER_M_DEFAULT,
+            minimum=0,
+            help="Fallback only; used when the embeddings upstream returns no native cost.",
+            group="Pricing",
+        ),
         ConfigField(
             "gateway_max_retries",
             "Max retries",
@@ -290,6 +331,11 @@ class GatewayService(BaseService):
         cfg["gateway_vision_key"] = cfg["gateway_vision_key"] or os.environ.get(
             "OPENROUTER_API_KEY", ""
         )
+        cfg["gateway_embed_key"] = (
+            cfg["gateway_embed_key"]
+            or cfg["gateway_vision_key"]
+            or os.environ.get("OPENROUTER_API_KEY", "")
+        )
         return cfg
 
     def authorize(self, request: Request) -> bool:
@@ -352,6 +398,18 @@ class GatewayService(BaseService):
                 self.log("Rejected chat request: JSON body was not an object")
                 raise HTTPException(status_code=400, detail="Invalid JSON body")
             return await runtime.handle_chat(body, cfg, owner, user_agent, self.log)
+        if subpath == "embeddings" and request.method == "POST":
+            try:
+                body = await request.json()
+            except Exception:
+                self.log("Rejected embeddings request: invalid JSON body")
+                raise HTTPException(status_code=400, detail="Invalid JSON body")
+            if not isinstance(body, dict):
+                self.log("Rejected embeddings request: JSON body was not an object")
+                raise HTTPException(status_code=400, detail="Invalid JSON body")
+            return await runtime.handle_embeddings(
+                body, cfg, owner, user_agent, self.log
+            )
         body = await request.body()
         content_type = request.headers.get("content-type", "")
         return await runtime.handle_passthrough(
@@ -395,6 +453,7 @@ class GatewayService(BaseService):
                 "in_flight": 0,
                 "peak_in_flight": 0,
                 "vision_calls": 0,
+                "embed_calls": 0,
                 "last_status": 0,
                 "last_latency_ms": 0,
                 "pool": 0,
@@ -406,11 +465,13 @@ class GatewayService(BaseService):
             {"label": "Requests (lifetime)", "value": m["requests"]},
             {"label": "In flight", "value": m["in_flight"]},
             {"label": "Vision calls", "value": m["vision_calls"]},
+            {"label": "Embed calls", "value": m["embed_calls"]},
             {"label": "Last status", "value": m["last_status"] or "-"},
             {"label": "Last latency", "value": f"{m['last_latency_ms']} ms"},
             {"label": "Pool size", "value": m["pool"]},
             {"label": "Circuit", "value": "open" if m["circuit_open"] else "closed"},
             {"label": "Model", "value": cfg["gateway_model"]},
+            {"label": "Embed model", "value": cfg["gateway_embed_model"]},
             {"label": "Requests 24h", "value": s["requests"]},
             {"label": "Success 24h", "value": f"{s['success_pct']}%"},
             {"label": "Error rate 24h", "value": f"{s['error_pct']}%"},