Feat: Add New API model provider for OpenAI-compatible gateways (#15991)

## Summary

Add support for **"New API"** as a model provider, enabling connection
to [New API](https://github.com/QuantumNous/new-api) /
[one-api](https://github.com/songquanpeng/one-api) compatible gateways
that aggregate multiple LLM backends behind a unified OpenAI-compatible
`/v1` endpoint.

### Features

- **All model types**: Chat, Embedding, Rerank, Image2Text, TTS,
Speech2Text
- **List Models discovery**: `NewAPI(OpenAIAPICompatible)` class in
`model_meta.py` queries the gateway's `/v1/models` to auto-discover
available models via the native `GET /api/v1/providers/<name>/models`
endpoint
- **Model parameter editing**: Pencil icon on each discovered model row
to edit `model_type`, `max_tokens`, and `features` (e.g. tool call
support) before submitting
- **Custom model addition**: "Add Custom Model" button at the bottom of
the List Models dropdown for models not returned by the API
- **Gear icon settings**: Enabled the Settings gear button on provider
instances to manage models on existing instances (viewMode)
- **viewMode credential passthrough**: Fixed List Models in viewMode —
merges `initialValues` credentials when `api_key`/`base_url` fields are
hidden by `hideWhenInstanceExists`

### Changes

**Backend** (8 files):
- `rag/llm/chat_model.py` — `NewAPIChat(Base)` class
- `rag/llm/embedding_model.py` — `NewAPIEmbed(OpenAIEmbed)` class (no
auto `/v1` append)
- `rag/llm/rerank_model.py` — `NewAPIRerank(Base)` class (uses `/rerank`
endpoint)
- `rag/llm/cv_model.py` — `NewAPICv(GptV4)` class
- `rag/llm/tts_model.py` — `NewAPITTS(OpenAITTS)` class
- `rag/llm/sequence2txt_model.py` — `NewAPISeq2txt(GPTSeq2txt)` class
- `rag/llm/model_meta.py` — `NewAPI(OpenAIAPICompatible)` class for List
Models discovery
- `conf/llm_factories.json` — New API factory entry with all model type
tags

**Frontend** (8 files + 1 new SVG):
- `web/src/assets/svg/llm/new-api.svg` — New API logo icon
- `web/src/constants/llm.ts` — `LLMFactory.NewAPI` enum + `IconMap`
entry
- `web/src/components/svg-icon.tsx` — `NewAPI` added to `svgIcons`
-
`web/src/pages/user-setting/setting-model/modal/provider-modal/field-config/local-llm-configs.ts`
— New API `buildLocalConfig`
-
`web/src/pages/user-setting/setting-model/modal/provider-modal/constants.ts`
— `LIST_MODEL_PROVIDERS` includes NewAPI
- `web/src/pages/user-setting/setting-model/components/used-model.tsx` —
Enable Settings gear button
-
`web/src/pages/user-setting/setting-model/modal/provider-modal/hooks/use-list-models-picker.ts`
— viewMode credential merge + model editing state/handlers
-
`web/src/pages/user-setting/setting-model/modal/provider-modal/hooks/use-list-models-options.tsx`
— Pencil edit icon per model row
-
`web/src/pages/user-setting/setting-model/modal/provider-modal/index.tsx`
— `AddCustomModelDialog` import + edit dialog rendering

**Note on Go implementation**: A Go model driver (`NewAPIModel`
delegating to `OpenAIModel`) has been prepared but is deferred until the
Go runtime is enabled in a future release (current v0.26.0 images use
`API_PROXY_SCHEME=python` and do not compile Go binaries). Will submit
as a follow-up PR.

## Related

- Depends on: #15996 (provider instance API improvements — server-side
credential lookup, idempotent `add_model`, security fixes — required for
viewMode gear icon and batch model submission)

## Test plan

- [ ] Add New API provider with api_key and base_url pointing to an
OpenAI-compatible gateway
- [ ] Click "List Models" — should discover and display available models
from `/v1/models`
- [ ] Click pencil icon on a model — should open edit dialog to change
model_type, max_tokens, features
- [ ] Select multiple models and click OK — should add all selected
models
- [ ] Click gear icon on the added instance — should open viewMode with
List Models working
- [ ] In viewMode, select new models including pre-existing ones, click
OK — should succeed (requires #15996)
- [ ] Verify all model types work: create a Chat assistant, Embedding
KB, Rerank setting

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Co-authored-by: Tim Wang <wanghualoong@users.noreply.github.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

conf/llm_factories.json

@@ -8058,6 +8058,14 @@
                     "model_type": "embedding"
                 }
             ]
+        },
+        {
+            "name": "New API",
+            "logo": "",
+            "tags": "LLM,TEXT EMBEDDING,TEXT RE-RANK,IMAGE2TEXT,TTS,SPEECH2TEXT",
+            "status": "1",
+            "llm": [],
+            "rank": "885"
         }
     ]
 }

rag/llm/chat_model.py

@@ -2100,3 +2100,13 @@ class RAGconChat(Base):
             base_url = "https://connect.ragcon.com/v1"
 
         super().__init__(key, model_name, base_url, **kwargs)
+
+
+class NewAPIChat(Base):
+    _FACTORY_NAME = "New API"
+
+    def __init__(self, key, model_name, base_url, **kwargs):
+        if not base_url:
+            raise ValueError("url cannot be None")
+        model_name = model_name.split("___")[0]
+        super().__init__(key, model_name, base_url, **kwargs)

rag/llm/cv_model.py

@@ -1417,4 +1417,17 @@ class BedrockCV(Base):
         return res.choices[0].message.content.strip(), total_token_count_from_response(res)
 
     def describe(self, image):
-        return self.describe_with_prompt(image)
+        return self.describe_with_prompt(image)
+
+
+class NewAPICv(GptV4):
+    _FACTORY_NAME = "New API"
+
+    def __init__(self, key, model_name, lang="Chinese", base_url="", **kwargs):
+        if not base_url:
+            raise ValueError("url cannot be None")
+        self.client = OpenAI(api_key=key, base_url=base_url)
+        self.async_client = AsyncOpenAI(api_key=key, base_url=base_url)
+        self.model_name = model_name.split("___")[0]
+        self.lang = lang
+        Base.__init__(self, **kwargs)

rag/llm/embedding_model.py

@@ -1300,3 +1300,13 @@ class PerplexityEmbed(Base):
     def encode_queries(self, text):
         embds, cnt = self.encode([text])
         return np.array(embds[0]), cnt
+
+
+class NewAPIEmbed(OpenAIEmbed):
+    _FACTORY_NAME = "New API"
+
+    def __init__(self, key, model_name, base_url):
+        if not base_url:
+            raise ValueError("url cannot be None")
+        self.client = OpenAI(api_key=key, base_url=base_url)
+        self.model_name = model_name.split("___")[0]

rag/llm/model_meta.py

@@ -469,3 +469,16 @@ class VLLM(OpenAIAPICompatible):
 
 class LMStudio(OpenAIAPICompatible):
     _FACTORY_NAME = "LM-Studio"
+
+
+class NewAPI(OpenAIAPICompatible):
+    _FACTORY_NAME = "New API"
+
+    def _get_api_key(self):
+        try:
+            parsed = json.loads(self.api_key)
+            if isinstance(parsed, dict):
+                return parsed.get("api_key", self.api_key)
+        except (JSONDecodeError, TypeError):
+            pass
+        return self.api_key

rag/llm/rerank_model.py

@@ -623,3 +623,37 @@ class RAGconRerank(Base):
         except Exception as _e:
             log_exception(_e, res)
         return rank, token_count
+
+
+class NewAPIRerank(Base):
+    _FACTORY_NAME = "New API"
+
+    def __init__(self, key, model_name, base_url):
+        normalized_base_url = (base_url or "").strip()
+        if "/rerank" in normalized_base_url:
+            self.base_url = normalized_base_url.rstrip("/")
+        else:
+            self.base_url = urljoin(f"{normalized_base_url.rstrip('/')}/", "rerank").rstrip("/")
+        self.headers = {
+            "Content-Type": "application/json",
+            "Authorization": f"Bearer {key}",
+        }
+        self.model_name = model_name.split("___")[0]
+
+    def _compute_rank(self, query: str, texts: list):
+        texts = [truncate(t, 500) for t in texts]
+        data = {
+            "model": self.model_name,
+            "query": query,
+            "documents": texts,
+            "top_n": len(texts),
+        }
+        token_count = sum(num_tokens_from_string(t) for t in texts)
+        res = requests.post(self.base_url, headers=self.headers, json=data).json()
+        rank = np.zeros(len(texts), dtype=float)
+        try:
+            for d in res["results"]:
+                rank[d["index"]] = d["relevance_score"]
+        except Exception as _e:
+            log_exception(_e, res)
+        return rank, token_count

rag/llm/sequence2txt_model.py

@@ -433,3 +433,13 @@ class RAGconSeq2txt(Base):
         # Return text and token count
         text = transcription.text.strip()
         return text, num_tokens_from_string(text)
+
+
+class NewAPISeq2txt(GPTSeq2txt):
+    _FACTORY_NAME = "New API"
+
+    def __init__(self, key, model_name="whisper-1", base_url="", **kwargs):
+        if not base_url:
+            raise ValueError("url cannot be None")
+        model_name = model_name.split("___")[0]
+        super().__init__(key, model_name=model_name, base_url=base_url, **kwargs)

rag/llm/tts_model.py

@@ -543,3 +543,13 @@ class RAGconTTS(Base):
         for chunk in response.iter_content(chunk_size=1024):
             if chunk:
                 yield chunk
+
+
+class NewAPITTS(OpenAITTS):
+    _FACTORY_NAME = "New API"
+
+    def __init__(self, key, model_name, base_url="", **kwargs):
+        if not base_url:
+            raise ValueError("url cannot be None")
+        model_name = model_name.split("___")[0]
+        super().__init__(key, model_name, base_url, **kwargs)

web/src/components/svg-icon.tsx

@@ -88,6 +88,7 @@ const svgIcons = [
   // LLMFactory.DeerAPI,
   LLMFactory.Avian,
   LLMFactory.RAGcon,
+  LLMFactory.NewAPI,
 ];
 
 export const LlmIcon = ({

web/src/constants/llm.ts

@@ -72,6 +72,7 @@ export enum LLMFactory {
   Avian = 'Avian',
   RAGcon = 'RAGcon',
   Perplexity = 'Perplexity',
+  NewAPI = 'New API',
 }
 
 // Please lowercase the file name
@@ -143,6 +144,7 @@ export const IconMap = {
   [LLMFactory.Avian]: 'avian',
   [LLMFactory.RAGcon]: 'ragcon',
   [LLMFactory.Perplexity]: 'perplexity',
+  [LLMFactory.NewAPI]: 'new-api',
 };
 
 export const ModelTypeToField: Record<string, string> = {

web/src/pages/user-setting/setting-model/components/used-model.tsx

@@ -265,11 +265,9 @@ function InstanceModelList({
               </span>
             ))}
           </div>
-          {false && (
-            <Button size="icon" variant="ghost" onClick={handleSettingsClick}>
-              <Settings size={12} />
-            </Button>
-          )}
+          <Button size="icon" variant="ghost" onClick={handleSettingsClick}>
+            <Settings size={12} />
+          </Button>
         </div>
       )}
 

web/src/pages/user-setting/setting-model/modal/provider-modal/constants.ts

@@ -23,6 +23,7 @@ export const LIST_MODEL_PROVIDERS = new Set<string>([
   LLMFactory.Xinference,
   LLMFactory.LocalAI,
   LLMFactory.BaiduYiYan,
+  LLMFactory.NewAPI,
 
   // LLMFactory.HuggingFace,
   // LLMFactory.GoogleCloud,

web/src/pages/user-setting/setting-model/modal/provider-modal/field-config/local-llm-configs.ts

@@ -134,6 +134,15 @@ export const LocalLlmConfigs: Record<string, ProviderConfig> = {
     undefined,
     'https://docs.vllm.ai/en/latest/',
   ),
+  [LLMFactory.NewAPI]: buildLocalConfig(
+    LLMFactory.NewAPI,
+    'New API',
+    ['chat', 'embedding', 'rerank', 'image2text', 'tts', 'speech2text'],
+    undefined,
+    false,
+    undefined,
+    'https://github.com/QuantumNous/new-api',
+  ),
   // [LLMFactory.TokenPony]: buildLocalConfig(
   //   LLMFactory.TokenPony,
   //   'TokenPony',

web/src/pages/user-setting/setting-model/modal/provider-modal/hooks/use-list-models-options.tsx

@@ -1,6 +1,7 @@
 import { Checkbox } from '@/components/ui/checkbox';
 import { useTranslate } from '@/hooks/common-hooks';
 import { IProviderModelItem } from '@/interfaces/request/llm';
+import { Pencil } from 'lucide-react';
 import { useMemo } from 'react';
 
 interface UseListModelsOptionsParams {
@@ -9,29 +10,16 @@ interface UseListModelsOptionsParams {
   allSelected: boolean;
   handleSelectModel: (model: IProviderModelItem) => void;
   handleToggleAll: () => void;
+  onEditModel?: (model: IProviderModelItem) => void;
 }
 
-/**
- * Build ToggleList options from the fetched model list. The first item is
- * a sentinel "All models" row that toggles the full selection.
- *
- * Why the Checkbox uses `onClick` (not `onCheckedChange`):
- * Radix Checkbox calls `event.stopPropagation()` internally on its onClick
- * when the Checkbox lives inside a form, so the row's `onClick` (attached
- * by ToggleList) never fires when the user clicks the Checkbox itself.
- * To make the Checkbox click toggle selection, we handle it in our own
- * `onClick` and re-stop propagation to (a) prevent the row's onClick from
- * double-firing and (b) make Radix's CheckboxBubbleInput dispatch a
- * non-bubbling synthetic click on its hidden form input — without this,
- * the dispatched click would bubble back to the row and re-trigger the
- * toggle, causing "Maximum update depth exceeded".
- */
 export const useListModelsOptions = ({
   models,
   selectedModelItems,
   allSelected,
   handleSelectModel,
   handleToggleAll,
+  onEditModel,
 }: UseListModelsOptionsParams) => {
   const { t } = useTranslate('setting');
 
@@ -75,13 +63,29 @@ export const useListModelsOptions = ({
                   );
                 })}
             </div>
-            <Checkbox
-              checked={checked}
-              onClick={(e) => {
-                e.stopPropagation();
-                handleSelectModel(m);
-              }}
-            />
+            <div className="flex items-center gap-1">
+              {onEditModel && (
+                <button
+                  type="button"
+                  aria-label="Edit model"
+                  title="Edit model"
+                  className="p-1 rounded hover:bg-bg-card text-text-secondary"
+                  onClick={(e) => {
+                    e.stopPropagation();
+                    onEditModel(m);
+                  }}
+                >
+                  <Pencil size={12} />
+                </button>
+              )}
+              <Checkbox
+                checked={checked}
+                onClick={(e) => {
+                  e.stopPropagation();
+                  handleSelectModel(m);
+                }}
+              />
+            </div>
           </div>
         ),
         onClick: () => handleSelectModel(m),
@@ -98,6 +102,7 @@ export const useListModelsOptions = ({
     handleSelectModel,
     allSelected,
     handleToggleAll,
+    onEditModel,
     t,
   ]);
 };

web/src/pages/user-setting/setting-model/modal/provider-modal/hooks/use-list-models-picker.ts

@@ -154,15 +154,16 @@ export const useListModelsPicker = ({
       if (models.length > 0 || listLoading) return;
       setListLoading(true);
       try {
-        const values = (formRef.current?.getValues() || {}) as Record<
+        const rawValues = (formRef.current?.getValues() || {}) as Record<
           string,
           any
         >;
-        // Reuse the verifyTransform to build the request payload — it
-        // already knows how to flatten provider-specific auth (api_key,
-        // base_url, region, model_info). Pass the current modelInfoList
-        // (empty on first load, populated on re-opens) so the backend
-        // sees an array shape consistent with the verify/submit payloads.
+        // In viewMode, pass instance_name so the backend looks up
+        // stored credentials server-side (avoids exposing api_key).
+        // For new-instance mode, pass api_key/base_url from the form.
+        const values = viewMode && initialValues
+          ? { ...initialValues, ...rawValues }
+          : rawValues;
         const verifyArgs = config.verifyTransform
           ? config.verifyTransform({
               ...values,
@@ -171,10 +172,13 @@ export const useListModelsPicker = ({
           : { apiKey: values.api_key ?? '', baseUrl: values.base_url };
         const res = await listProviderModels({
           provider_name: llmFactory,
-          api_key: (verifyArgs as any).apiKey ?? '',
-          base_url: (verifyArgs as any).baseUrl,
+          api_key: viewMode ? '' : ((verifyArgs as any).apiKey ?? ''),
+          base_url: viewMode ? '' : ((verifyArgs as any).baseUrl ?? ''),
           region: (verifyArgs as any).region,
           model_info: (verifyArgs as any).modelInfo ?? modelInfoList,
+          ...(viewMode && initialValues?.instance_name
+            ? { instance_name: initialValues.instance_name }
+            : {}),
         });
         if (res?.code === 0 && Array.isArray(res.data)) {
           setModelsState(res.data);
@@ -210,6 +214,8 @@ export const useListModelsPicker = ({
       llmFactory,
       modelInfoList,
       formRef,
+      viewMode,
+      initialValues,
     ],
   );
 
@@ -287,6 +293,30 @@ export const useListModelsPicker = ({
     [],
   );
 
+  // --- Model editing ---
+  const [editingModel, setEditingModel] =
+    useState<IProviderModelItem | null>(null);
+  const [editDialogOpen, setEditDialogOpen] = useState(false);
+
+  const handleEditModel = useCallback((model: IProviderModelItem) => {
+    setEditingModel(model);
+    setEditDialogOpen(true);
+  }, []);
+
+  const handleSaveEditedModel = useCallback(
+    (updated: IProviderModelItem) => {
+      const oldName = editingModel?.name;
+      if (!oldName) return;
+      const replace = (items: IProviderModelItem[]) =>
+        items.map((m) => (m.name === oldName ? updated : m));
+      setModelsState(replace);
+      setSelectedModelItemsState(replace);
+      setEditDialogOpen(false);
+      setEditingModel(null);
+    },
+    [editingModel],
+  );
+
   return {
     models,
     listLoading,
@@ -298,5 +328,10 @@ export const useListModelsPicker = ({
     handleToggleAll,
     setModels,
     setSelectedModelItems,
+    editingModel,
+    editDialogOpen,
+    setEditDialogOpen,
+    handleEditModel,
+    handleSaveEditedModel,
   };
 };

web/src/pages/user-setting/setting-model/modal/provider-modal/index.tsx

@@ -14,6 +14,7 @@ import { FieldValues } from 'react-hook-form';
 import { LLMHeader } from '../../components/llm-header';
 import VerifyButton from '../verify-button';
 import { AddableToggleList } from './components/addable-toggle-list';
+import { AddCustomModelDialog } from './components/add-custom-model-dialog';
 import { useCustomModelFields } from './components/use-custom-model-fields';
 import {
   useListModelsOptions,
@@ -103,6 +104,11 @@ const ProviderModal = ({
     handleSelectModel,
     handleToggleAll,
     setModels,
+    editingModel,
+    editDialogOpen,
+    setEditDialogOpen,
+    handleEditModel,
+    handleSaveEditedModel,
   } = useListModelsPicker({
     visible,
     hasModelNameField,
@@ -164,6 +170,7 @@ const ProviderModal = ({
     allSelected,
     handleSelectModel,
     handleToggleAll,
+    onEditModel: handleEditModel,
   });
 
   // Submit and verify handlers — branch on viewMode and on whether the
@@ -270,6 +277,32 @@ const ProviderModal = ({
           </div>
         </div>
       </DynamicForm.Root>
+      {editingModel && (
+        <AddCustomModelDialog
+          open={editDialogOpen}
+          onOpenChange={setEditDialogOpen}
+          title={editingModel.name}
+          fields={customModelDialogFields.map((f) => ({
+            ...f,
+            defaultValue:
+              f.name === 'name'
+                ? editingModel.name
+                : f.name === 'model_types'
+                  ? editingModel.model_types ?? []
+                  : f.name === 'max_tokens'
+                    ? editingModel.max_tokens ?? 0
+                    : f.name === 'features'
+                      ? editingModel.features ?? []
+                      : f.defaultValue,
+          }))}
+          onSubmit={handleSaveEditedModel}
+          submitText={tc('ok')}
+          cancelText={tc('cancel')}
+          existingNames={existingNames.filter(
+            (n) => n !== editingModel.name,
+          )}
+        />
+      )}
     </Modal>
   );
 };