Merge 937e2b5a54 into 995bef73de

add: model gpt-5

.env.template

@@ -64,6 +64,11 @@ CUSTOM_MODELS=
 # Change default model
 DEFAULT_MODEL=
 
+# (optional)
+# Default: Empty
+# Change default compress model
+DEFAULT_COMPRESS_MODEL=
+
 # anthropic claude Api Key.(optional)
 ANTHROPIC_API_KEY=
 

README.md

@@ -328,6 +328,10 @@ For ByteDance: use `modelName@bytedance=deploymentName` to customize model name
 
 Change default model
 
+### `DEFAULT_COMPRESS_MODEL` （optional）
+
+Change default summary model
+
 ### `VISION_MODELS` (optional)
 
 > Default: Empty

README_CN.md

@@ -253,6 +253,10 @@ DeepSeek Api Url.
 
 更改默认模型
 
+### `DEFAULT_COMPRESS_MODEL` （可选）
+
+更改默认对话摘要模型
+
 ### `VISION_MODELS` (可选)
 
 > 默认值：空

README_JA.md

@@ -221,6 +221,10 @@ ByteDance モードでは、`modelName@bytedance=deploymentName` 形式でモデ
 
 デフォルトのモデルを変更します。
 
+### `DEFAULT_COMPRESS_MODEL` （オプション）
+
+デフォルトの圧縮モデルを変更します。
+
 ### `VISION_MODELS` (オプション)
 
 > デフォルト：空

app/api/config/route.ts

@@ -14,6 +14,7 @@ const DANGER_CONFIG = {
   disableFastLink: serverConfig.disableFastLink,
   customModels: serverConfig.customModels,
   defaultModel: serverConfig.defaultModel,
+  defaultCompressModel: serverConfig.defaultCompressModel,
   visionModels: serverConfig.visionModels,
 };
 

app/client/platforms/openai.ts

@@ -200,6 +200,7 @@ export class ChatGPTApi implements LLMApi {
       options.config.model.startsWith("o1") ||
       options.config.model.startsWith("o3") ||
       options.config.model.startsWith("o4-mini");
+    const isGpt5 =  options.config.model.startsWith("gpt-5");
     if (isDalle3) {
       const prompt = getMessageTextContent(
         options.messages.slice(-1)?.pop() as any,
@@ -230,7 +231,7 @@ export class ChatGPTApi implements LLMApi {
         messages,
         stream: options.config.stream,
         model: modelConfig.model,
-        temperature: !isO1OrO3 ? modelConfig.temperature : 1,
+        temperature: (!isO1OrO3 && !isGpt5) ? modelConfig.temperature : 1,
         presence_penalty: !isO1OrO3 ? modelConfig.presence_penalty : 0,
         frequency_penalty: !isO1OrO3 ? modelConfig.frequency_penalty : 0,
         top_p: !isO1OrO3 ? modelConfig.top_p : 1,
@@ -238,7 +239,13 @@ export class ChatGPTApi implements LLMApi {
         // Please do not ask me why not send max_tokens, no reason, this param is just shit, I dont want to explain anymore.
       };
 
-      if (isO1OrO3) {
+      if (isGpt5) {
+  	// Remove max_tokens if present
+  	delete requestPayload.max_tokens;
+  	// Add max_completion_tokens (or max_completion_tokens if that's what you meant)
+  	requestPayload["max_completion_tokens"] = modelConfig.max_tokens;
+
+      } else if (isO1OrO3) {
         // by default the o1/o3 models will not attempt to produce output that includes markdown formatting
         // manually add "Formatting re-enabled" developer message to encourage markdown inclusion in model responses
         // (https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/reasoning?tabs=python-secure#markdown-output)
@@ -251,8 +258,9 @@ export class ChatGPTApi implements LLMApi {
         requestPayload["max_completion_tokens"] = modelConfig.max_tokens;
       }
 
+
       // add max_tokens to vision model
-      if (visionModel && !isO1OrO3) {
+      if (visionModel && !isO1OrO3 && ! isGpt5) {
         requestPayload["max_tokens"] = Math.max(modelConfig.max_tokens, 4000);
       }
     }

app/components/model-config.tsx

@@ -259,13 +259,15 @@ export function ModelConfigList(props: {
             });
           }}
         >
-          {allModels
-            .filter((v) => v.available)
-            .map((v, i) => (
-              <option value={`${v.name}@${v.provider?.providerName}`} key={i}>
-                {v.displayName}({v.provider?.providerName})
-              </option>
-            ))}
+          {Object.keys(groupModels).map((providerName, index) => (
+            <optgroup label={providerName} key={index}>
+              {groupModels[providerName].map((v, i) => (
+                <option value={`${v.name}@${v.provider?.providerName}`} key={i}>
+                  {v.displayName}
+                </option>
+              ))}
+            </optgroup>
+          ))}
         </Select>
       </ListItem>
     </>

app/config/server.ts

@@ -23,6 +23,7 @@ declare global {
       DISABLE_FAST_LINK?: string; // disallow parse settings from url or not
       CUSTOM_MODELS?: string; // to control custom models
       DEFAULT_MODEL?: string; // to control default model in every new chat window
+      DEFAULT_COMPRESS_MODEL?: string; // to control default compress model
       VISION_MODELS?: string; // to control vision models
 
       // stability only
@@ -139,6 +140,7 @@ export const getServerSideConfig = () => {
   const disableGPT4 = !!process.env.DISABLE_GPT4;
   let customModels = process.env.CUSTOM_MODELS ?? "";
   let defaultModel = process.env.DEFAULT_MODEL ?? "";
+  let defaultCompressModel = process.env.DEFAULT_COMPRESS_MODEL ?? "";
   let visionModels = process.env.VISION_MODELS ?? "";
 
   if (disableGPT4) {
@@ -149,6 +151,9 @@ export const getServerSideConfig = () => {
     if (defaultModel && isGPT4Model(defaultModel)) {
       defaultModel = "";
     }
+    if (defaultCompressModel && isGPT4Model(defaultCompressModel)) {
+      defaultCompressModel = "";
+    }
   }
 
   const isStability = !!process.env.STABILITY_API_KEY;
@@ -271,6 +276,7 @@ export const getServerSideConfig = () => {
     disableFastLink: !!process.env.DISABLE_FAST_LINK,
     customModels,
     defaultModel,
+    defaultCompressModel,
     visionModels,
     allowedWebDavEndpoints,
     enableMcp: process.env.ENABLE_MCP === "true",

app/constant.ts

@@ -493,6 +493,7 @@ export const VISION_MODEL_REGEXES = [
   /o3/,
   /o4-mini/,
   /grok-4/i,
+  /gpt-5/
 ];
 
 export const EXCLUDE_VISION_MODEL_REGEXES = [/claude-3-5-haiku-20241022/];
@@ -517,6 +518,11 @@ const openaiModels = [
   "gpt-4.1-nano-2025-04-14",
   "gpt-4.5-preview",
   "gpt-4.5-preview-2025-02-27",
+  "gpt-5-chat",
+  "gpt-5-mini",
+  "gpt-5-nano",
+  "gpt-5",
+  "gpt-5-chat-2025-01-01-preview",
   "gpt-4o",
   "gpt-4o-2024-05-13",
   "gpt-4o-2024-08-06",

app/store/access.ts

@@ -147,6 +147,7 @@ const DEFAULT_ACCESS_STATE = {
   disableFastLink: false,
   customModels: "",
   defaultModel: "",
+  defaultCompressModel: "",
   visionModels: "",
 
   // tts config
@@ -262,12 +263,21 @@ export const useAccessStore = createPersistStore(
         .then((res) => res.json())
         .then((res) => {
           const defaultModel = res.defaultModel ?? "";
-          if (defaultModel !== "") {
+          if (defaultModel) {
             const [model, providerName] = getModelProvider(defaultModel);
             DEFAULT_CONFIG.modelConfig.model = model;
             DEFAULT_CONFIG.modelConfig.providerName = providerName as any;
           }
 
+          const defaultCompressModel = res.defaultCompressModel ?? "";
+          if (defaultCompressModel) {
+            const [model, providerName] =
+              getModelProvider(defaultCompressModel);
+            DEFAULT_CONFIG.modelConfig.compressModel = model;
+            DEFAULT_CONFIG.modelConfig.compressProviderName =
+              providerName as any;
+          }
+
           return res;
         })
         .then((res: DangerConfig) => {

app/store/chat.ts

@@ -123,9 +123,16 @@ function getSummarizeModel(
   currentModel: string,
   providerName: string,
 ): string[] {
+  const configStore = useAppConfig.getState();
+  if (configStore.modelConfig.compressModel) {
+    return [
+      configStore.modelConfig.compressModel,
+      configStore.modelConfig.compressProviderName,
+    ];
+  }
+
   // if it is using gpt-* models, force to use 4o-mini to summarize
   if (currentModel.startsWith("gpt") || currentModel.startsWith("chatgpt")) {
-    const configStore = useAppConfig.getState();
     const accessStore = useAccessStore.getState();
     const allModel = collectModelsWithDefaultModel(
       configStore.models,