diff --git a/app/client/platforms/openai.ts b/app/client/platforms/openai.ts
index 919716bfb..629158843 100644
--- a/app/client/platforms/openai.ts
+++ b/app/client/platforms/openai.ts
@@ -110,6 +110,16 @@ 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.
     };
 
+    // add max_tokens to vision model
+    if (visionModel) {
+      Object.defineProperty(requestPayload, "max_tokens", {
+        enumerable: true,
+        configurable: true,
+        writable: true,
+        value: modelConfig.max_tokens,
+      });
+    }
+
     console.log("[Request] openai payload: ", requestPayload);
 
     const shouldStream = !!options.config.stream;
diff --git a/app/components/chat.tsx b/app/components/chat.tsx
index 369d84c33..bcd0e605d 100644
--- a/app/components/chat.tsx
+++ b/app/components/chat.tsx
@@ -6,6 +6,7 @@ import React, {
   useMemo,
   useCallback,
   Fragment,
+  RefObject,
 } from "react";
 
 import SendWhiteIcon from "../icons/send-white.svg";
@@ -382,11 +383,13 @@ function ChatAction(props: {
   );
 }
 
-function useScrollToBottom() {
+function useScrollToBottom(
+  scrollRef: RefObject<HTMLDivElement>,
+  detach: boolean = false,
+) {
   // for auto-scroll
-  const scrollRef = useRef<HTMLDivElement>(null);
-  const [autoScroll, setAutoScroll] = useState(true);
 
+  const [autoScroll, setAutoScroll] = useState(true);
   function scrollDomToBottom() {
     const dom = scrollRef.current;
     if (dom) {
@@ -399,7 +402,7 @@ function useScrollToBottom() {
 
   // auto scroll
   useEffect(() => {
-    if (autoScroll) {
+    if (autoScroll && !detach) {
       scrollDomToBottom();
     }
   });
@@ -658,7 +661,17 @@ function _Chat() {
   const [userInput, setUserInput] = useState("");
   const [isLoading, setIsLoading] = useState(false);
   const { submitKey, shouldSubmit } = useSubmitHandler();
-  const { scrollRef, setAutoScroll, scrollDomToBottom } = useScrollToBottom();
+  const scrollRef = useRef<HTMLDivElement>(null);
+  const isScrolledToBottom = scrollRef?.current
+    ? Math.abs(
+        scrollRef.current.scrollHeight -
+          (scrollRef.current.scrollTop + scrollRef.current.clientHeight),
+      ) <= 1
+    : false;
+  const { setAutoScroll, scrollDomToBottom } = useScrollToBottom(
+    scrollRef,
+    isScrolledToBottom,
+  );
   const [hitBottom, setHitBottom] = useState(true);
   const isMobileScreen = useMobileScreen();
   const navigate = useNavigate();
@@ -1003,7 +1016,6 @@ function _Chat() {
     setHitBottom(isHitBottom);
     setAutoScroll(isHitBottom);
   };
-
   function scrollToBottom() {
     setMsgRenderIndex(renderMessages.length - CHAT_PAGE_SIZE);
     scrollDomToBottom();
@@ -1088,6 +1100,47 @@ function _Chat() {
     };
     // eslint-disable-next-line react-hooks/exhaustive-deps
   }, []);
+  
+  const handlePaste = useCallback(
+    async (event: React.ClipboardEvent<HTMLTextAreaElement>) => {
+      const currentModel = chatStore.currentSession().mask.modelConfig.model;
+      if(!isVisionModel(currentModel)){return;}
+      const items = (event.clipboardData || window.clipboardData).items;
+      for (const item of items) {
+        if (item.kind === "file" && item.type.startsWith("image/")) {
+          event.preventDefault();
+          const file = item.getAsFile();
+          if (file) {
+            const images: string[] = [];
+            images.push(...attachImages);
+            images.push(
+              ...(await new Promise<string[]>((res, rej) => {
+                setUploading(true);
+                const imagesData: string[] = [];
+                compressImage(file, 256 * 1024)
+                  .then((dataUrl) => {
+                    imagesData.push(dataUrl);
+                    setUploading(false);
+                    res(imagesData);
+                  })
+                  .catch((e) => {
+                    setUploading(false);
+                    rej(e);
+                  });
+              })),
+            );
+            const imagesLength = images.length;
+
+            if (imagesLength > 3) {
+              images.splice(3, imagesLength - 3);
+            }
+            setAttachImages(images);
+          }
+        }
+      }
+    },
+    [attachImages, chatStore],
+  );
 
   async function uploadImage() {
     const images: string[] = [];
@@ -1437,6 +1490,7 @@ function _Chat() {
             onKeyDown={onInputKeyDown}
             onFocus={scrollToBottom}
             onClick={scrollToBottom}
+            onPaste={handlePaste}
             rows={inputRows}
             autoFocus={autoFocus}
             style={{
diff --git a/app/utils.ts b/app/utils.ts
index 33b8eccd2..b484e8386 100644
--- a/app/utils.ts
+++ b/app/utils.ts
@@ -9,8 +9,9 @@ export function trimTopic(topic: string) {
   // This will remove the specified punctuation from the end of the string
   // and also trim quotes from both the start and end if they exist.
   return topic
-    .replace(/^["“”]+|["“”]+$/g, "")
-    .replace(/[，。！？”“"、,.!?]*$/, "");
+    // fix for gemini
+    .replace(/^["“”*]+|["“”*]+$/g, "")
+    .replace(/[，。！？”“"、,.!?*]*$/, "");
 }
 
 export async function copyToClipboard(text: string) {
@@ -292,8 +293,8 @@ export function getMessageImages(message: RequestMessage): string[] {
 
 export function isVisionModel(model: string) {
   return (
-    model.startsWith("gpt-4-vision") ||
-    model.startsWith("gemini-pro-vision") ||
-    !DEFAULT_MODELS.find((m) => m.name == model)
+    // model.startsWith("gpt-4-vision") ||
+    // model.startsWith("gemini-pro-vision") ||
+    model.includes("vision")
   );
 }