feat: add placeholder model metas (#15753)

### What problem does this PR solve?

add placeholder model metas

### Type of change

- [x] Bug Fix (non-breaking change which fixes an issue)

api/apps/services/provider_api_service.py

@@ -184,7 +184,14 @@ async def list_provider_models(provider_name: str, api_key: str = None, base_url
             "name": llm["name"],
             "max_tokens": llm["max_tokens"],
             "model_types": [llm["model_type"]],
-            "features": None
+            "features": (
+                llm.get("features")
+                if llm.get("features") is not None
+                else (
+                    (["is_tools"] if llm.get("is_tools") else [])
+                    + (["thinking"] if llm.get("thinking") else [])
+                )
+            )
         } for llm in factory_info[0]["llm"]]
 
     model_base_url = base_url or factory_info[0].get("url", "")

conf/llm_factories.json

@@ -3916,7 +3916,56 @@
             "logo": "",
             "tags": "LLM,TTS",
             "status": "1",
-            "llm": []
+            "llm": [
+                {
+                    "llm_name": "Spark-Max",
+                    "tags": "LLM,CHAT,8K",
+                    "max_tokens": 8192,
+                    "model_type": "chat",
+                    "is_tools": true,
+                    "features": ["is_tools", "thinking"]
+                },
+                {
+                    "llm_name": "Spark-Max-32K",
+                    "tags": "LLM,CHAT,32K",
+                    "max_tokens": 32768,
+                    "model_type": "chat",
+                    "is_tools": true,
+                    "features": ["is_tools", "thinking"]
+                },
+                {
+                    "llm_name": "Spark-Lite",
+                    "tags": "LLM,CHAT,8K",
+                    "max_tokens": 8192,
+                    "model_type": "chat",
+                    "is_tools": true,
+                    "features": ["is_tools", "thinking"]
+                },
+                {
+                    "llm_name": "Spark-Pro",
+                    "tags": "LLM,CHAT,8K",
+                    "max_tokens": 8192,
+                    "model_type": "chat",
+                    "is_tools": true,
+                    "features": ["is_tools", "thinking"]
+                },
+                {
+                    "llm_name": "Spark-Pro-128K",
+                    "tags": "LLM,CHAT,128K",
+                    "max_tokens": 131072,
+                    "model_type": "chat",
+                    "is_tools": true,
+                    "features": ["is_tools", "thinking"]
+                },
+                {
+                    "llm_name": "Spark-4.0-Ultra",
+                    "tags": "LLM,CHAT,128K",
+                    "max_tokens": 131072,
+                    "model_type": "chat",
+                    "is_tools": true,
+                    "features": ["is_tools", "thinking"]
+                }
+            ]
         },
         {
             "name": "BaiduYiyan",

rag/llm/model_meta.py

@@ -13,6 +13,7 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import json
 import aiohttp
 from abc import ABC
 
@@ -110,3 +111,107 @@ class Ollama(Base):
                         "max_tokens": model_info["model_info"].get(max_tokens_key, 8192)
                     })
         return res
+
+
+class FishAudio(Base):
+    _FACTORY_NAME = "Fish Audio"
+
+    def _get_access_token(self):
+        api_key = self._get_api_key()
+        if not api_key:
+            return ""
+        try:
+            payload = json.loads(api_key)
+        except Exception:
+            return api_key
+        if isinstance(payload, dict):
+            return payload.get("fish_audio_ak") or payload.get("access_token") or payload.get("api_key") or api_key
+        return api_key
+
+    def _get_model_list_url(self):
+        if not self.base_url:
+            return "https://api.fish.audio/model"
+        base_url = self.base_url.rstrip("/")
+        if "/v1/" in base_url:
+            return base_url.split("/v1")[0].rstrip("/") + "/model"
+        if base_url.endswith("/v1"):
+            return base_url[:-3] + "/model"
+        return base_url + "/model"
+
+    async def get_model_list(self):
+        url = self._get_model_list_url()
+        access_token = self._get_access_token()
+        if not url or not access_token:
+            return []
+
+        async with aiohttp.ClientSession() as session:
+            async with session.get(url, headers={"Authorization": f"Bearer {access_token}"}) as resp:
+                if resp.status != 200:
+                    return []
+                raw_model_list = await resp.json()
+                if not isinstance(raw_model_list, dict):
+                    return []
+                models = raw_model_list.get("items") or []
+                if not isinstance(models, list):
+                    return []
+
+                model_list = []
+                for model in models:
+                    if not isinstance(model, dict):
+                        continue
+                    model_name = model.get("title") or model.get("_id")
+                    if not model_name:
+                        continue
+                    model_list.append({
+                        "name": model_name,
+                        "model_types": [LLMType.TTS.value],
+                        "features": [],
+                        "max_tokens": 8192,
+                    })
+                return model_list
+
+class MinerU(Base):
+    _FACTORY_NAME = "MinerU"
+
+    def _get_access_token(self):
+        api_key = self._get_api_key()
+        if not api_key:
+            return ""
+        try:
+            payload = json.loads(api_key)
+        except Exception:
+            return api_key
+        if isinstance(payload, dict):
+            return payload.get("access_token") or payload.get("api_key") or api_key
+        return api_key
+
+    async def get_model_list(self):
+        url = self._get_model_list_url()
+        access_token = self._get_access_token()
+        if not url or not access_token:
+            return []
+
+        async with aiohttp.ClientSession() as session:
+            async with session.get(url, headers={"Authorization": f"Bearer {access_token}"}) as resp:
+                if resp.status != 200:
+                    return []
+                raw_model_list = await resp.json()
+                if isinstance(raw_model_list, dict):
+                    raw_model_list = raw_model_list.get("data") or raw_model_list.get("models") or raw_model_list.get("items") or []
+                if not isinstance(raw_model_list, list):
+                    return []
+
+                model_list = []
+                for model in raw_model_list:
+                    if not isinstance(model, dict):
+                        continue
+                    model_name = model.get("title") or model.get("name") or model.get("id") or model.get("_id")
+                    if not model_name:
+                        continue
+                    model_list.append({
+                        "name": model_name,
+                        "model_types": [LLMType.OCR.value],
+                        "features": [],
+                        "max_tokens": model.get("max_tokens", 8192),
+                    })
+                return model_list