send longer chat history

.gitignore

@@ -49,3 +49,4 @@ masks.json
 
 # mcp config
 app/mcp/mcp_config.json
+Dockerfile.local

app/client/platforms/anthropic.ts

@@ -71,8 +71,6 @@ const ClaudeMapper = {
   system: "user",
 } as const;
 
-const keys = ["claude-2, claude-instant-1"];
-
 export class ClaudeApi implements LLMApi {
   speech(options: SpeechOptions): Promise<ArrayBuffer> {
     throw new Error("Method not implemented.");

app/client/platforms/google.ts

@@ -197,8 +197,6 @@ export class GeminiProApi implements LLMApi {
         signal: controller.signal,
         headers: getHeaders(),
       };
-
-      const isThinking = options.config.model.includes("-thinking");
       // make a fetch request
       const requestTimeoutId = setTimeout(
         () => controller.abort(),

app/client/platforms/openai.ts

@@ -67,6 +67,8 @@ export interface RequestPayload {
   top_p: number;
   max_tokens?: number;
   max_completion_tokens?: number;
+  reasoning_effort?: string;
+  // O3 only
 }
 
 export interface DalleRequestPayload {
@@ -196,9 +198,9 @@ export class ChatGPTApi implements LLMApi {
     let requestPayload: RequestPayload | DalleRequestPayload;
 
     const isDalle3 = _isDalle3(options.config.model);
-    const isO1OrO3 =
-      options.config.model.startsWith("o1") ||
-      options.config.model.startsWith("o3");
+    const isO1 = options.config.model.startsWith("o1");
+    const isO3 = options.config.model.startsWith("o3");
+    const isO1OrO3 = isO1 || isO3;
     if (isDalle3) {
       const prompt = getMessageTextContent(
         options.messages.slice(-1)?.pop() as any,
@@ -242,9 +244,18 @@ export class ChatGPTApi implements LLMApi {
         requestPayload["max_completion_tokens"] = modelConfig.max_tokens;
       }
 
+      if (isO3) {
+        requestPayload["reasoning_effort"] = "high";
+        // make o3-mini defaults to high reasoning effort
+      }
+
       // add max_tokens to vision model
       if (visionModel) {
-        requestPayload["max_tokens"] = Math.max(modelConfig.max_tokens, 4000);
+        if (isO1) {
+          requestPayload["max_completion_tokens"] = modelConfig.max_tokens;
+        } else {
+          requestPayload["max_tokens"] = Math.max(modelConfig.max_tokens, 4000);
+        }
       }
     }
 
@@ -286,6 +297,11 @@ export class ChatGPTApi implements LLMApi {
           isDalle3 ? OpenaiPath.ImagePath : OpenaiPath.ChatPath,
         );
       }
+      // make a fetch request
+      const requestTimeoutId = setTimeout(
+        () => controller.abort(),
+        getTimeoutMSByModel(options.config.model),
+      );
       if (shouldStream) {
         let index = -1;
         const [tools, funcs] = usePluginStore
@@ -393,12 +409,6 @@ export class ChatGPTApi implements LLMApi {
           headers: getHeaders(),
         };
 
-        // make a fetch request
-        const requestTimeoutId = setTimeout(
-          () => controller.abort(),
-          getTimeoutMSByModel(options.config.model),
-        );
-
         const res = await fetch(chatPath, chatPayload);
         clearTimeout(requestTimeoutId);
 

app/components/chat.tsx

@@ -18,6 +18,7 @@ import ReturnIcon from "../icons/return.svg";
 import CopyIcon from "../icons/copy.svg";
 import SpeakIcon from "../icons/speak.svg";
 import SpeakStopIcon from "../icons/speak-stop.svg";
+import LoadingIcon from "../icons/three-dots.svg";
 import LoadingButtonIcon from "../icons/loading.svg";
 import PromptIcon from "../icons/prompt.svg";
 import MaskIcon from "../icons/mask.svg";
@@ -78,6 +79,8 @@ import {
 
 import { uploadImage as uploadImageRemote } from "@/app/utils/chat";
 
+import dynamic from "next/dynamic";
+
 import { ChatControllerPool } from "../client/controller";
 import { DalleQuality, DalleStyle, ModelSize } from "../typing";
 import { Prompt, usePromptStore } from "../store/prompt";
@@ -122,15 +125,14 @@ import { getModelProvider } from "../utils/model";
 import { RealtimeChat } from "@/app/components/realtime-chat";
 import clsx from "clsx";
 import { getAvailableClientsCount, isMcpEnabled } from "../mcp/actions";
-import { Markdown } from "./markdown";
 
 const localStorage = safeLocalStorage();
 
 const ttsPlayer = createTTSPlayer();
 
-// const Markdown = dynamic(async () => (await import("./markdown")).Markdown, {
-//   loading: () => <LoadingIcon />,
-// });
+const Markdown = dynamic(async () => (await import("./markdown")).Markdown, {
+  loading: () => <LoadingIcon />,
+});
 
 const MCPAction = () => {
   const navigate = useNavigate();
@@ -1982,8 +1984,6 @@ function _Chat() {
                               fontFamily={fontFamily}
                               parentRef={scrollRef}
                               defaultShow={i >= messages.length - 6}
-                              immediatelyRender={i >= messages.length - 3}
-                              streaming={message.streaming}
                             />
                             {getMessageImages(message).length == 1 && (
                               <img

app/components/markdown.tsx

@@ -267,136 +267,6 @@ function tryWrapHtmlCode(text: string) {
     );
 }
 
-// Split content into paragraphs while preserving code blocks
-function splitContentIntoParagraphs(content: string) {
-  // Check for unclosed code blocks
-  const codeBlockStartCount = (content.match(/```/g) || []).length;
-  let processedContent = content;
-
-  // Add closing tag if there's an odd number of code block markers
-  if (codeBlockStartCount % 2 !== 0) {
-    processedContent = content + "\n```";
-  }
-
-  // Extract code blocks
-  const codeBlockRegex = /```[\s\S]*?```/g;
-  const codeBlocks: string[] = [];
-  let codeBlockCounter = 0;
-
-  // Replace code blocks with placeholders
-  const contentWithPlaceholders = processedContent.replace(
-    codeBlockRegex,
-    (match) => {
-      codeBlocks.push(match);
-      const placeholder = `__CODE_BLOCK_${codeBlockCounter++}__`;
-      return placeholder;
-    },
-  );
-
-  // Split by double newlines
-  const paragraphs = contentWithPlaceholders
-    .split(/\n\n+/)
-    .filter((p) => p.trim());
-
-  // Restore code blocks
-  return paragraphs.map((p) => {
-    if (p.match(/__CODE_BLOCK_\d+__/)) {
-      return p.replace(/__CODE_BLOCK_\d+__/g, (match) => {
-        const index = parseInt(match.match(/\d+/)?.[0] || "0");
-        return codeBlocks[index] || match;
-      });
-    }
-    return p;
-  });
-}
-
-// Lazy-loaded paragraph component
-function MarkdownParagraph({
-  content,
-  onLoad,
-}: {
-  content: string;
-  onLoad?: () => void;
-}) {
-  const [isLoaded, setIsLoaded] = useState(false);
-  const placeholderRef = useRef<HTMLDivElement>(null);
-  const [isVisible, setIsVisible] = useState(false);
-
-  useEffect(() => {
-    let observer: IntersectionObserver;
-    if (placeholderRef.current) {
-      observer = new IntersectionObserver(
-        (entries) => {
-          if (entries[0].isIntersecting) {
-            setIsVisible(true);
-          }
-        },
-        { threshold: 0.1, rootMargin: "200px 0px" },
-      );
-      observer.observe(placeholderRef.current);
-    }
-    return () => observer?.disconnect();
-  }, []);
-
-  useEffect(() => {
-    if (isVisible && !isLoaded) {
-      setIsLoaded(true);
-      onLoad?.();
-    }
-  }, [isVisible, isLoaded, onLoad]);
-
-  // Generate preview content
-  const previewContent = useMemo(() => {
-    if (content.startsWith("```")) {
-      return "```" + (content.split("\n")[0] || "").slice(3) + "...```";
-    }
-    return content.length > 60 ? content.slice(0, 60) + "..." : content;
-  }, [content]);
-
-  return (
-    <div className="markdown-paragraph" ref={placeholderRef}>
-      {!isLoaded ? (
-        <div className="markdown-paragraph-placeholder">{previewContent}</div>
-      ) : (
-        <_MarkDownContent content={content} />
-      )}
-    </div>
-  );
-}
-
-// Memoized paragraph component to prevent unnecessary re-renders
-const MemoizedMarkdownParagraph = React.memo(
-  ({ content }: { content: string }) => {
-    return <_MarkDownContent content={content} />;
-  },
-  (prevProps, nextProps) => prevProps.content === nextProps.content,
-);
-
-MemoizedMarkdownParagraph.displayName = "MemoizedMarkdownParagraph";
-
-// Specialized component for streaming content
-function StreamingMarkdownContent({ content }: { content: string }) {
-  const paragraphs = useMemo(
-    () => splitContentIntoParagraphs(content),
-    [content],
-  );
-  const lastParagraphRef = useRef<HTMLDivElement>(null);
-
-  return (
-    <div className="markdown-streaming-content">
-      {paragraphs.map((paragraph, index) => (
-        <div
-          key={`p-${index}-${paragraph.substring(0, 20)}`}
-          className="markdown-paragraph markdown-streaming-paragraph"
-          ref={index === paragraphs.length - 1 ? lastParagraphRef : null}
-        >
-          <MemoizedMarkdownParagraph content={paragraph} />
-        </div>
-      ))}
-    </div>
-  );
-}
-
 function _MarkDownContent(props: { content: string }) {
   const escapedContent = useMemo(() => {
     return tryWrapHtmlCode(escapeBrackets(props.content));
@@ -456,27 +326,9 @@ export function Markdown(
     fontFamily?: string;
     parentRef?: RefObject<HTMLDivElement>;
     defaultShow?: boolean;
-    immediatelyRender?: boolean;
-    streaming?: boolean; // Whether this is a streaming response
   } & React.DOMAttributes<HTMLDivElement>,
 ) {
   const mdRef = useRef<HTMLDivElement>(null);
-  const paragraphs = useMemo(
-    () => splitContentIntoParagraphs(props.content),
-    [props.content],
-  );
-  const [loadedCount, setLoadedCount] = useState(0);
-
-  // Determine rendering strategy based on props
-  const shouldAsyncRender =
-    !props.immediatelyRender && !props.streaming && paragraphs.length > 1;
-
-  useEffect(() => {
-    // Immediately render all paragraphs if specified
-    if (props.immediatelyRender) {
-      setLoadedCount(paragraphs.length);
-    }
-  }, [props.immediatelyRender, paragraphs.length]);
 
   return (
     <div
@@ -492,24 +344,6 @@ export function Markdown(
     >
       {props.loading ? (
         <LoadingIcon />
-      ) : props.streaming ? (
-        // Use specialized component for streaming content
-        <StreamingMarkdownContent content={props.content} />
-      ) : shouldAsyncRender ? (
-        <div className="markdown-content">
-          {paragraphs.map((paragraph, index) => (
-            <MarkdownParagraph
-              key={index}
-              content={paragraph}
-              onLoad={() => setLoadedCount((prev) => prev + 1)}
-            />
-          ))}
-          {loadedCount < paragraphs.length && loadedCount > 0 && (
-            <div className="markdown-paragraph-loading">
-              <LoadingIcon />
-            </div>
-          )}
-        </div>
       ) : (
         <MarkdownContent content={props.content} />
       )}

app/constant.ts

@@ -408,35 +408,20 @@ You are an AI assistant with access to system tools. Your role is to help users
    
 `;
 
-export const SUMMARIZE_MODEL = "gpt-4o-mini";
-export const GEMINI_SUMMARIZE_MODEL = "gemini-pro";
+export const SUMMARIZE_MODEL = "gpt-4.1-mini";
+export const GEMINI_SUMMARIZE_MODEL = "gemini-2.0-flash";
 export const DEEPSEEK_SUMMARIZE_MODEL = "deepseek-chat";
 
 export const KnowledgeCutOffDate: Record<string, string> = {
-  default: "2021-09",
-  "gpt-4-turbo": "2023-12",
-  "gpt-4-turbo-2024-04-09": "2023-12",
-  "gpt-4-turbo-preview": "2023-12",
-  "gpt-4o": "2023-10",
-  "gpt-4o-2024-05-13": "2023-10",
-  "gpt-4o-2024-08-06": "2023-10",
-  "gpt-4o-2024-11-20": "2023-10",
-  "chatgpt-4o-latest": "2023-10",
-  "gpt-4o-mini": "2023-10",
-  "gpt-4o-mini-2024-07-18": "2023-10",
-  "gpt-4-vision-preview": "2023-04",
-  "o1-mini-2024-09-12": "2023-10",
-  "o1-mini": "2023-10",
-  "o1-preview-2024-09-12": "2023-10",
-  "o1-preview": "2023-10",
-  "o1-2024-12-17": "2023-10",
-  o1: "2023-10",
-  "o3-mini-2025-01-31": "2023-10",
-  "o3-mini": "2023-10",
+  default: "2023-10",
   // After improvements,
   // it's now easier to add "KnowledgeCutOffDate" instead of stupid hardcoding it, as was done previously.
-  "gemini-pro": "2023-12",
-  "gemini-pro-vision": "2023-12",
+  "gemini-2.5-pro-exp-03-25": "2025-01",
+  "gemini-2.0-flash": "2024-08",
+  "claude-3-7-sonnet-latest": "2024-10",
+  "claude-3-5-haiku-latest": "2024-10",
+  "gpt-4.1": "2024-06",
+  "gpt-4.1-mini": "2024-06",
   "deepseek-chat": "2024-07",
   "deepseek-coder": "2024-07",
 };
@@ -457,11 +442,12 @@ export const DEFAULT_TTS_VOICES = [
 
 export const VISION_MODEL_REGEXES = [
   /vision/,
-  /gpt-4o/,
+  /gpt-4\.1/,
   /claude-3/,
   /gemini-1\.5/,
   /gemini-exp/,
   /gemini-2\.0/,
+  /gemini-2\.5-pro/,
   /learnlm/,
   /qwen-vl/,
   /qwen2-vl/,
@@ -469,78 +455,23 @@ export const VISION_MODEL_REGEXES = [
   /^dall-e-3$/, // Matches exactly "dall-e-3"
   /glm-4v/,
   /vl/i,
+  /o1/,
 ];
 
 export const EXCLUDE_VISION_MODEL_REGEXES = [/claude-3-5-haiku-20241022/];
 
-const openaiModels = [
-  // As of July 2024, gpt-4o-mini should be used in place of gpt-3.5-turbo,
-  // as it is cheaper, more capable, multimodal, and just as fast. gpt-3.5-turbo is still available for use in the API.
-  "gpt-3.5-turbo",
-  "gpt-3.5-turbo-1106",
-  "gpt-3.5-turbo-0125",
-  "gpt-4",
-  "gpt-4-0613",
-  "gpt-4-32k",
-  "gpt-4-32k-0613",
-  "gpt-4-turbo",
-  "gpt-4-turbo-preview",
-  "gpt-4o",
-  "gpt-4o-2024-05-13",
-  "gpt-4o-2024-08-06",
-  "gpt-4o-2024-11-20",
-  "chatgpt-4o-latest",
-  "gpt-4o-mini",
-  "gpt-4o-mini-2024-07-18",
-  "gpt-4-vision-preview",
-  "gpt-4-turbo-2024-04-09",
-  "gpt-4-1106-preview",
-  "dall-e-3",
-  "o1-mini",
-  "o1-preview",
-  "o3-mini",
-];
+const openaiModels = ["dall-e-3", "o1", "o3-mini", "gpt-4.1", "gpt-4.1-mini"];
 
 const googleModels = [
-  "gemini-1.0-pro", // Deprecated on 2/15/2025
-  "gemini-1.5-pro-latest",
-  "gemini-1.5-pro",
-  "gemini-1.5-pro-002",
-  "gemini-1.5-pro-exp-0827",
-  "gemini-1.5-flash-latest",
-  "gemini-1.5-flash-8b-latest",
-  "gemini-1.5-flash",
-  "gemini-1.5-flash-8b",
-  "gemini-1.5-flash-002",
-  "gemini-1.5-flash-exp-0827",
-  "learnlm-1.5-pro-experimental",
-  "gemini-exp-1114",
-  "gemini-exp-1121",
-  "gemini-exp-1206",
   "gemini-2.0-flash",
-  "gemini-2.0-flash-exp",
-  "gemini-2.0-flash-lite-preview-02-05",
-  "gemini-2.0-flash-thinking-exp",
-  "gemini-2.0-flash-thinking-exp-1219",
-  "gemini-2.0-flash-thinking-exp-01-21",
-  "gemini-2.0-pro-exp",
-  "gemini-2.0-pro-exp-02-05",
+  "gemini-2.0-flash-lite",
+  "gemini-2.5-pro-exp-03-25",
 ];
 
 const anthropicModels = [
-  "claude-instant-1.2",
-  "claude-2.0",
-  "claude-2.1",
-  "claude-3-sonnet-20240229",
-  "claude-3-opus-20240229",
   "claude-3-opus-latest",
-  "claude-3-haiku-20240307",
-  "claude-3-5-haiku-20241022",
   "claude-3-5-haiku-latest",
-  "claude-3-5-sonnet-20240620",
-  "claude-3-5-sonnet-20241022",
   "claude-3-5-sonnet-latest",
-  "claude-3-7-sonnet-20250219",
   "claude-3-7-sonnet-latest",
 ];
 

app/store/config.ts

@@ -66,14 +66,14 @@ export const DEFAULT_CONFIG = {
   modelConfig: {
     model: "gpt-4o-mini" as ModelType,
     providerName: "OpenAI" as ServiceProvider,
-    temperature: 0.5,
+    temperature: 0.2,
     top_p: 1,
     max_tokens: 4000,
     presence_penalty: 0,
     frequency_penalty: 0,
     sendMemory: true,
-    historyMessageCount: 4,
-    compressMessageLengthThreshold: 1000,
+    historyMessageCount: 20,
+    compressMessageLengthThreshold: 5000,
     compressModel: "",
     compressProviderName: "",
     enableInjectSystemPrompts: true,

app/styles/markdown.scss

@@ -99,7 +99,6 @@
   font-size: 14px;
   line-height: 1.5;
   word-wrap: break-word;
-  margin-bottom: 0;
 }
 
 .light {
@@ -359,14 +358,8 @@
 .markdown-body kbd {
   display: inline-block;
   padding: 3px 5px;
-  font:
-    11px ui-monospace,
-    SFMono-Regular,
-    SF Mono,
-    Menlo,
-    Consolas,
-    Liberation Mono,
-    monospace;
+  font: 11px ui-monospace, SFMono-Regular, SF Mono, Menlo, Consolas,
+    Liberation Mono, monospace;
   line-height: 10px;
   color: var(--color-fg-default);
   vertical-align: middle;
@@ -455,28 +448,16 @@
 .markdown-body tt,
 .markdown-body code,
 .markdown-body samp {
-  font-family:
-    ui-monospace,
-    SFMono-Regular,
-    SF Mono,
-    Menlo,
-    Consolas,
-    Liberation Mono,
-    monospace;
+  font-family: ui-monospace, SFMono-Regular, SF Mono, Menlo, Consolas,
+    Liberation Mono, monospace;
   font-size: 12px;
 }
 
 .markdown-body pre {
   margin-top: 0;
   margin-bottom: 0;
-  font-family:
-    ui-monospace,
-    SFMono-Regular,
-    SF Mono,
-    Menlo,
-    Consolas,
-    Liberation Mono,
-    monospace;
+  font-family: ui-monospace, SFMono-Regular, SF Mono, Menlo, Consolas,
+    Liberation Mono, monospace;
   font-size: 12px;
   word-wrap: normal;
 }
@@ -1149,87 +1130,3 @@
 #dmermaid {
   display: none;
 }
-
-.markdown-content {
-  width: 100%;
-}
-
-.markdown-paragraph {
-  transition: opacity 0.3s ease;
-  margin-bottom: 0.5em;
-
-  &.markdown-paragraph-visible {
-    opacity: 1;
-  }
-
-  &.markdown-paragraph-hidden {
-    opacity: 0.7;
-  }
-}
-
-.markdown-paragraph-placeholder {
-  padding: 8px;
-  color: var(--color-fg-subtle);
-  background-color: var(--color-canvas-subtle);
-  border-radius: 6px;
-  border-left: 3px solid var(--color-border-muted);
-  white-space: nowrap;
-  overflow: hidden;
-  text-overflow: ellipsis;
-  font-family: var(--font-family-sans);
-  font-size: 14px;
-  min-height: 1.2em;
-}
-
-.markdown-paragraph-loading {
-  height: 20px;
-  background-color: var(--color-canvas-subtle);
-  border-radius: 6px;
-  margin-bottom: 8px;
-  position: relative;
-  overflow: hidden;
-
-  &::after {
-    content: "";
-    position: absolute;
-    top: 0;
-    left: 0;
-    width: 30%;
-    height: 100%;
-    background: linear-gradient(
-      90deg,
-      transparent,
-      rgba(255, 255, 255, 0.1),
-      transparent
-    );
-    animation: shimmer 1.5s infinite;
-  }
-}
-
-@keyframes shimmer {
-  0% {
-    transform: translateX(-100%);
-  }
-  100% {
-    transform: translateX(200%);
-  }
-}
-
-.markdown-streaming-content {
-  width: 100%;
-}
-
-.markdown-streaming-paragraph {
-  opacity: 1;
-  animation: fadeIn 0.3s ease-in-out;
-  margin-bottom: 0.5em;
-}
-
-@keyframes fadeIn {
-  from {
-    opacity: 0.5;
-  }
-  to {
-    opacity: 1;
-  }
-}

app/utils.ts

@@ -304,9 +304,18 @@ export function getTimeoutMSByModel(model: string) {
     model.startsWith("o1") ||
     model.startsWith("o3") ||
     model.includes("deepseek-r") ||
-    model.includes("-thinking")
-  )
+    model.includes("-thinking") ||
+    model.includes("pro")
+  ) {
+    console.log(
+      "thinking model is " +
+        model +
+        " timeout is " +
+        REQUEST_TIMEOUT_MS_FOR_THINKING,
+    );
     return REQUEST_TIMEOUT_MS_FOR_THINKING;
+  }
+  console.log("normal model is " + model + " timeout is " + REQUEST_TIMEOUT_MS);
   return REQUEST_TIMEOUT_MS;
 }
 

test/model-available.test.ts

@@ -3,7 +3,7 @@ import { isModelNotavailableInServer } from "../app/utils/model";
 describe("isModelNotavailableInServer", () => {
   test("test model will return false, which means the model is available", () => {
     const customModels = "";
-    const modelName = "gpt-4";
+    const modelName = "gpt-4.1";
     const providerNames = "OpenAI";
     const result = isModelNotavailableInServer(
       customModels,

test/vision-model-checker.test.ts

@@ -15,10 +15,11 @@ describe("isVisionModel", () => {
 
   test("should identify vision models using regex patterns", () => {
     const visionModels = [
-      "gpt-4-vision",
+      "gpt-4.1",
       "claude-3-opus",
       "gemini-1.5-pro",
       "gemini-2.0",
+      "gemini-2.5-pro",
       "gemini-exp-vision",
       "learnlm-vision",
       "qwen-vl-max",