fix: model arguments fixed

backend/app/core/llm_provider.py

@@ -2,9 +2,10 @@ from typing import Optional, Dict
 
 from nanobot.providers.azure_openai_provider import AzureOpenAIProvider
 from nanobot.providers.openai_codex_provider import OpenAICodexProvider
-from nanobot.providers.openai_compat_provider import OpenAICompatProvider
 from nanobot.providers.registry import find_by_name
 
+from app.core.patched_openai_compat_provider import PatchedOpenAICompatProvider
+
 
 def normalize_provider_name(provider: Optional[str]) -> Optional[str]:
     if not provider:
@@ -51,7 +52,7 @@ def build_llm_provider(
             extra_headers=extra_headers,
         )
 
-    return OpenAICompatProvider(
+    return PatchedOpenAICompatProvider(
         api_key=api_key,
         api_base=api_base,
         default_model=model,

backend/app/core/nanobot.py

@@ -19,7 +19,6 @@ from nanobot.bus.events import OutboundMessage
 from nanobot.bus.queue import MessageBus
 from nanobot.config.loader import load_config
 from nanobot.cron.service import CronService
-from nanobot.providers.openai_compat_provider import OpenAICompatProvider
 from nanobot.providers.openai_codex_provider import OpenAICodexProvider
 from nanobot.providers.azure_openai_provider import AzureOpenAIProvider
 from nanobot.providers.base import GenerationSettings
@@ -32,6 +31,7 @@ from nanobot.config.schema import Config
 # or just import here if we are confident.
 # Given the structure, importing here should be fine as long as skills.py doesn't import nanobot.py.
 from app.api.skills import load_skills
+from app.core.patched_openai_compat_provider import PatchedOpenAICompatProvider
 from app.services.llm_cache import get_llm_configs, get_active_llm_config
 
 from app.core.data_root import get_workspace_root
@@ -45,6 +45,7 @@ class NanobotIntegration:
         self._started = False
         self._model_agent_cache: Dict[tuple[str | None, int | None], AgentLoop] = {}
         self._model_agent_lock = asyncio.Lock()
+        self._last_usage_by_session: Dict[str, Dict[str, Any]] = {}
 
     @staticmethod
     def _normalize_config_value(value: Any) -> Any:
@@ -80,6 +81,28 @@ class NanobotIntegration:
             return content
         return str(response)
 
+    @staticmethod
+    def _normalize_usage(usage: Any) -> Dict[str, int] | None:
+        if not isinstance(usage, dict):
+            return None
+        normalized: Dict[str, int] = {}
+        prompt = int(usage.get("prompt_tokens", 0) or 0)
+        completion = int(usage.get("completion_tokens", 0) or 0)
+        total = int(usage.get("total_tokens", 0) or 0)
+        
+        # If total_tokens is missing or zero, calculate it
+        if total == 0:
+            total = prompt + completion
+            
+        normalized["prompt_tokens"] = prompt
+        normalized["completion_tokens"] = completion
+        normalized["total_tokens"] = total
+        return normalized if (prompt > 0 or completion > 0) else None
+
+    def get_last_usage(self, session_id: str) -> Dict[str, int] | None:
+        usage = self._last_usage_by_session.get(session_id)
+        return dict(usage) if usage else None
+
     def _need_custom_agent_for_target(self, target_config: Dict[str, Any]) -> bool:
         if not self.agent:
             return False
@@ -220,7 +243,7 @@ class NanobotIntegration:
                 extra_headers=extra_headers,
             )
 
-        return OpenAICompatProvider(
+        return PatchedOpenAICompatProvider(
             api_key=api_key,
             api_base=api_base,
             default_model=model,
@@ -407,6 +430,9 @@ class NanobotIntegration:
             on_progress=on_progress,
             on_stream=on_stream,
         )
+        usage = self._normalize_usage(getattr(agent_to_use, "_last_usage", None))
+        if usage:
+            self._last_usage_by_session[session_id] = usage
         return self._extract_response_text(response)
 
     def _normalize_session_messages(self, messages: List[Any]) -> List[dict[str, Any]]:

backend/app/core/patched_openai_compat_provider.py

@@ -0,0 +1,43 @@
+from __future__ import annotations
+
+from typing import Any
+
+from nanobot.providers.openai_compat_provider import OpenAICompatProvider
+
+
+class PatchedOpenAICompatProvider(OpenAICompatProvider):
+    _MAX_COMPLETION_TOKEN_MODELS = ("gpt-5", "o1", "o3", "o4")
+
+    def _build_kwargs(
+        self,
+        messages: list[dict[str, Any]],
+        tools: list[dict[str, Any]] | None,
+        model: str | None,
+        max_tokens: int,
+        temperature: float,
+        reasoning_effort: str | None,
+        tool_choice: str | dict[str, Any] | None,
+    ) -> dict[str, Any]:
+        kwargs = super()._build_kwargs(
+            messages=messages,
+            tools=tools,
+            model=model,
+            max_tokens=max_tokens,
+            temperature=temperature,
+            reasoning_effort=reasoning_effort,
+            tool_choice=tool_choice,
+        )
+
+        model_name = (model or self.default_model or "").lower()
+        spec = self._spec
+        supports_max_completion_tokens = bool(
+            spec and getattr(spec, "supports_max_completion_tokens", False)
+        )
+        should_use_max_completion_tokens = supports_max_completion_tokens or any(
+            token in model_name for token in self._MAX_COMPLETION_TOKEN_MODELS
+        )
+
+        if should_use_max_completion_tokens and "max_tokens" in kwargs:
+            kwargs["max_completion_tokens"] = kwargs.pop("max_tokens")
+
+        return kwargs

backend/main.py

@@ -268,6 +268,7 @@ def _persist_assistant_enrichment(
     session_id: str,
     viz_payload: Optional[Dict[str, Any]] = None,
     artifacts: Optional[List[Dict[str, Any]]] = None,
+    usage: Optional[Dict[str, Any]] = None,
 ) -> None:
     if not nanobot_service.agent:
         return
@@ -281,9 +282,25 @@ def _persist_assistant_enrichment(
     if artifacts:
         session.messages[-1]["artifacts"] = artifacts
         changed = True
+    if usage:
+        session.messages[-1]["usage"] = usage
+        changed = True
     if changed:
         nanobot_service.agent.sessions.save(session)
 
+
+def _extract_reasoning_content(session_messages: List[Dict[str, Any]]) -> str:
+    for message in reversed(session_messages):
+        if not isinstance(message, dict):
+            continue
+        if message.get("role") != "assistant":
+            continue
+        reasoning_content = message.get("reasoning_content")
+        if isinstance(reasoning_content, str) and reasoning_content.strip():
+            return reasoning_content
+        break
+    return ""
+
 @app.post("/nanobot/chat")
 async def nanobot_chat(request: ChatRequest):
     try:
@@ -321,10 +338,12 @@ async def nanobot_chat(request: ChatRequest):
         artifacts = extract_artifacts(text, session_messages)
 
         viz_payload = current_viz_data.get()
+        usage = nanobot_service.get_last_usage(request.session_id)
         _persist_assistant_enrichment(
             session_id=request.session_id,
             viz_payload=viz_payload if isinstance(viz_payload, dict) else None,
             artifacts=artifacts,
+            usage=usage,
         )
 
         payload = {
@@ -334,6 +353,8 @@ async def nanobot_chat(request: ChatRequest):
         }
         if artifacts:
             payload["artifacts"] = artifacts
+        if usage:
+            payload["usage"] = usage
         return payload
     except Exception as e:
         raise HTTPException(status_code=500, detail=str(e))
@@ -356,7 +377,9 @@ async def nanobot_chat_stream(request: ChatRequest):
 
             async def _on_progress(content: str, **kwargs: Any) -> None:
                 if content:
-                    await progress_queue.put(content)
+                    payload: Dict[str, Any] = {"type": "progress", "content": content}
+                    payload.update(kwargs)
+                    await progress_queue.put(payload)
 
             async def _on_stream(delta: str) -> None:
                 if delta:
@@ -427,9 +450,10 @@ async def nanobot_chat_stream(request: ChatRequest):
                 session = nanobot_service.agent.sessions.get_or_create(request.session_id)
                 session_messages = session.messages
             artifacts = extract_artifacts(text, session_messages)
+            reasoning_content = _extract_reasoning_content(session_messages)
 
-            # Check again for viz payload after task completes if not sent yet
             viz_payload = current_viz_data.get()
+            usage = nanobot_service.get_last_usage(request.session_id)
             if viz_payload:
                 try:
                     current_hash = hash((
@@ -447,11 +471,16 @@ async def nanobot_chat_stream(request: ChatRequest):
                 session_id=request.session_id,
                 viz_payload=viz_payload if isinstance(viz_payload, dict) else None,
                 artifacts=artifacts,
+                usage=usage,
             )
             
             final_payload = {"type": "final", "content": text}
+            if reasoning_content:
+                final_payload["reasoning_content"] = reasoning_content
             if artifacts:
                 final_payload["artifacts"] = artifacts
+            if usage:
+                final_payload["usage"] = usage
             yield f"data: {json.dumps(final_payload, ensure_ascii=False)}\n\n"
             yield f"data: {json.dumps({'type': 'done'}, ensure_ascii=False)}\n\n"
         except asyncio.CancelledError:

backend/tests/test_chat_project_id.py

@@ -98,3 +98,48 @@ def test_nanobot_chat_stream_syncs_project_id(monkeypatch) -> None:
     assert "stream-complete" in content
     assert calls == [{"session_key": "api:test-3", "project_id": 202}]
     assert process_kwargs and process_kwargs[0]["project_id"] == 202
+
+
+def test_nanobot_chat_stream_emits_reasoning_flags_and_final_reasoning(monkeypatch) -> None:
+    async def fake_process_message(*args, **kwargs):
+        on_progress = kwargs.get("on_progress")
+        on_stream = kwargs.get("on_stream")
+        if on_progress:
+            await on_progress("模型正在拆解问题", is_reasoning=True)
+            await on_progress("开始执行工具", tool_hint=True)
+        if on_stream:
+            await on_stream("answer-token")
+        return "final-answer"
+
+    class _DummySession:
+        def __init__(self):
+            self.metadata = {}
+            self.messages = [
+                {"role": "assistant", "content": "final-answer", "reasoning_content": "完整思考过程"}
+            ]
+
+    class _DummySessions:
+        def get_or_create(self, _key):
+            return _DummySession()
+
+    class _DummyAgent:
+        def __init__(self):
+            self.sessions = _DummySessions()
+
+    async def collect_stream_chunks(response) -> list[str]:
+        chunks: list[str] = []
+        async for chunk in response.body_iterator:
+            chunks.append(chunk.decode("utf-8") if isinstance(chunk, bytes) else chunk)
+        return chunks
+
+    monkeypatch.setattr(main.nanobot_service, "process_message", fake_process_message)
+    monkeypatch.setattr(main.nanobot_service, "agent", _DummyAgent())
+
+    request = main.ChatRequest(message="hello", session_id="api:test-4", project_id=303)
+    response = asyncio.run(main.nanobot_chat_stream(request))
+    chunks = asyncio.run(collect_stream_chunks(response))
+    content = "".join(chunks)
+
+    assert '"type": "progress", "content": "模型正在拆解问题", "is_reasoning": true' in content
+    assert '"type": "progress", "content": "开始执行工具", "tool_hint": true' in content
+    assert '"type": "final", "content": "final-answer", "reasoning_content": "完整思考过程"' in content

backend/tests/test_patched_openai_compat_provider.py

@@ -0,0 +1,62 @@
+import sys
+from pathlib import Path
+
+BACKEND_ROOT = Path(__file__).resolve().parents[1]
+REPO_ROOT = BACKEND_ROOT.parent
+NANOBOT_ROOT = REPO_ROOT / "nanobot"
+if str(BACKEND_ROOT) not in sys.path:
+    sys.path.insert(0, str(BACKEND_ROOT))
+if str(NANOBOT_ROOT) not in sys.path:
+    sys.path.insert(0, str(NANOBOT_ROOT))
+
+from app.core.llm_provider import build_llm_provider
+from app.core.nanobot import NanobotIntegration
+from app.core.patched_openai_compat_provider import PatchedOpenAICompatProvider
+
+
+def test_build_llm_provider_uses_max_completion_tokens_for_gpt5() -> None:
+    provider = build_llm_provider(
+        model="gpt-5.4-nano",
+        provider="openai",
+        api_key="test-key",
+        api_base="https://example.com/v1",
+    )
+
+    assert isinstance(provider, PatchedOpenAICompatProvider)
+    kwargs = provider._build_kwargs(
+        messages=[{"role": "user", "content": "hello"}],
+        tools=None,
+        model="gpt-5.4-nano",
+        max_tokens=5,
+        temperature=0,
+        reasoning_effort=None,
+        tool_choice=None,
+    )
+
+    assert kwargs["max_completion_tokens"] == 5
+    assert "max_tokens" not in kwargs
+
+
+def test_nanobot_provider_keeps_max_tokens_for_legacy_models() -> None:
+    integration = NanobotIntegration()
+    provider = integration._build_provider(
+        model="gpt-4o-mini",
+        provider_name="openai",
+        api_key="test-key",
+        api_base="https://example.com/v1",
+        extra_headers=None,
+    )
+
+    assert isinstance(provider, PatchedOpenAICompatProvider)
+    kwargs = provider._build_kwargs(
+        messages=[{"role": "user", "content": "hello"}],
+        tools=None,
+        model="gpt-4o-mini",
+        max_tokens=5,
+        temperature=0,
+        reasoning_effort=None,
+        tool_choice=None,
+    )
+
+    assert kwargs["max_tokens"] == 5
+    assert "max_completion_tokens" not in kwargs

frontend/src/components/ChatInterface.tsx

@@ -1,6 +1,6 @@
 import { useState, useRef, useEffect } from "react";
 import { ScrollArea } from "@/components/ui/scroll-area";
-import { User, Loader2, ArrowUp, ChevronDown, Check, Square, Plus, Database, Wand2, Zap, CheckCircle2, Table, XCircle, Settings, ExternalLink, FileText, Download, Eye } from "lucide-react";
+import { User, Loader2, ArrowUp, ChevronDown, Check, Square, Plus, Database, Wand2, Zap, CheckCircle2, Table, XCircle, Settings, ExternalLink, FileText, Download, Eye, Copy } from "lucide-react";
 import { api } from "@/lib/api";
 import { type ChartSpec } from "@/store/visualizationStore";
 import { Popover, PopoverContent, PopoverTrigger } from "@/components/ui/popover";
@@ -25,6 +25,11 @@ interface Message {
   progressLogs?: string[];
   routeInfo?: string;
   reasoningContent?: string;
+  usage?: {
+    prompt_tokens: number;
+    completion_tokens: number;
+    total_tokens: number;
+  };
   artifacts?: MessageArtifact[];
 }
 
@@ -183,6 +188,8 @@ export function ChatInterface() {
   const [selectedSkillIds, setSelectedSkillIds] = useState<string[]>([]);
   const [isMenuOpen, setIsMenuOpen] = useState(false);
   const [artifactPreview, setArtifactPreview] = useState<ArtifactPreviewTarget | null>(null);
+  const [collapsedThinkingByMessage, setCollapsedThinkingByMessage] = useState<Record<string, boolean>>({});
+  const [thinkingCopiedByMessage, setThinkingCopiedByMessage] = useState<Record<string, boolean>>({});
   const scrollRef = useRef<HTMLDivElement>(null);
   const location = useLocation();
   const { currentProject } = useProjectStore();
@@ -372,6 +379,8 @@ export function ChatInterface() {
                 role: m.role as 'user' | 'assistant',
                 content: cleanContent,
                 viz: m.viz ? buildMessageViz(m.viz) : undefined,
+                reasoningContent: typeof m.reasoning_content === "string" ? m.reasoning_content : undefined,
+                usage: m.usage,
                 artifacts: normalizeArtifacts(m.artifacts),
               };
             });
@@ -489,6 +498,22 @@ export function ChatInterface() {
 
   const selectedDataSourceName = availableDataSources.find(ds => ds.id === selectedDataSource)?.name || "";
   const selectedSkills = availableSkills.filter(skill => selectedSkillIds.includes(skill.id));
+  const isThinkingCollapsed = (messageId: string) => collapsedThinkingByMessage[messageId] ?? true;
+  const toggleThinkingCollapsed = (messageId: string) => {
+    setCollapsedThinkingByMessage((prev) => ({ ...prev, [messageId]: !(prev[messageId] ?? true) }));
+  };
+  const copyThinkingContent = async (messageId: string, content: string) => {
+    if (!content.trim()) return;
+    try {
+      await navigator.clipboard.writeText(content);
+      setThinkingCopiedByMessage((prev) => ({ ...prev, [messageId]: true }));
+      window.setTimeout(() => {
+        setThinkingCopiedByMessage((prev) => ({ ...prev, [messageId]: false }));
+      }, 1200);
+    } catch (e) {
+      console.error("Failed to copy thinking content", e);
+    }
+  };
 
   const renderActiveSelections = () => {
     if (!selectedDataSource && selectedSkills.length === 0) return null;
@@ -613,9 +638,7 @@ export function ChatInterface() {
           if (msg.id !== assistantId) return msg;
           
           if (isReasoningToken) {
-            // 对于流式推理内容，拼接而不是创建新条目
+            return msg;
-            const currentReasoning = msg.reasoningContent || "";
-            return { ...msg, reasoningContent: currentReasoning + text };
           } else {
             // 对于普通的阶段性日志，取消 8 条限制，允许滚动查看所有历史
             const current = msg.progressLogs || [];
@@ -673,6 +696,35 @@ export function ChatInterface() {
       let hasDonePayload = false;
       let rafPending = false;
       let renderedText = "";
+      let reasoningBuffer = "";
+      let reasoningRafPending = false;
+
+      const flushReasoning = (force = false) => {
+        if (!reasoningBuffer) return;
+        if (force) {
+          const content = reasoningBuffer;
+          reasoningBuffer = "";
+          setMessagesForSession(targetSessionKey, (prev) =>
+            prev.map((msg) =>
+              msg.id === assistantId ? { ...msg, reasoningContent: (msg.reasoningContent || "") + content } : msg
+            )
+          );
+          return;
+        }
+        if (reasoningRafPending) return;
+        reasoningRafPending = true;
+        requestAnimationFrame(() => {
+          reasoningRafPending = false;
+          if (!reasoningBuffer) return;
+          const content = reasoningBuffer;
+          reasoningBuffer = "";
+          setMessagesForSession(targetSessionKey, (prev) =>
+            prev.map((msg) =>
+              msg.id === assistantId ? { ...msg, reasoningContent: (msg.reasoningContent || "") + content } : msg
+            )
+          );
+        });
+      };
 
       const flushAssistant = (force = false) => {
         if (streamedText === renderedText && !force) return;
@@ -717,6 +769,7 @@ export function ChatInterface() {
             type: string;
             content?: string;
             is_reasoning?: boolean;
+            tool_hint?: boolean;
             sql?: string;
             result?: unknown;
             error?: string;
@@ -724,6 +777,12 @@ export function ChatInterface() {
             reason?: string;
             chart?: { chart_spec?: ChartSpec | null; reasoning?: string; can_visualize?: boolean; chart_type?: string } | null;
             artifacts?: unknown;
+            reasoning_content?: string;
+            usage?: {
+              prompt_tokens: number;
+              completion_tokens: number;
+              total_tokens: number;
+            };
           };
 
            if (payload.type === "delta" && payload.content) {
@@ -743,9 +802,13 @@ export function ChatInterface() {
           }
 
           if (payload.type === "progress" && payload.content) {
-            // 如果 progress 内容带有空格或者换行，并且不是典型的系统提示词，很可能这是 reasoning_content
+            if (payload.is_reasoning || payload.tool_hint) {
-            // 为了安全起见，我们在后端应该加上 is_reasoning 标记，这里我们通过启发式或者统一拼接
+              const nextLine = payload.content.endsWith("\n") ? payload.content : `${payload.content}\n`;
-            pushProgressLog(payload.content, payload.is_reasoning || false);
+              reasoningBuffer += nextLine;
+              flushReasoning(false);
+            } else {
+              pushProgressLog(payload.content, false);
+            }
           }
 
            if (payload.type === "final") {
@@ -753,12 +816,22 @@ export function ChatInterface() {
             if (typeof payload.content === "string") {
               streamedText = payload.content;
             }
+            if (typeof payload.reasoning_content === "string") {
+              reasoningBuffer = "";
+              setMessagesForSession(targetSessionKey, (prev) =>
+                prev.map((msg) =>
+                  msg.id === assistantId ? { ...msg, reasoningContent: payload.reasoning_content } : msg
+                )
+              );
+            } else {
+              flushReasoning(true);
+            }
             flushAssistant(true);
             pushProgressLog(t('answerGenerationCompleted'));
             const messageArtifacts = normalizeArtifacts(payload.artifacts);
              setMessagesForSession(targetSessionKey, (prev) =>
                prev.map((msg) =>
-                msg.id === assistantId ? { ...msg, content: typeof payload.content === "string" ? payload.content : msg.content || "", awaitingFirstToken: false, viz: streamedViz ?? msg.viz, artifacts: messageArtifacts.length > 0 ? messageArtifacts : msg.artifacts } : msg
+                msg.id === assistantId ? { ...msg, content: typeof payload.content === "string" ? payload.content : msg.content || "", awaitingFirstToken: false, viz: streamedViz ?? msg.viz, usage: payload.usage, artifacts: messageArtifacts.length > 0 ? messageArtifacts : msg.artifacts } : msg
                )
              );
            }
@@ -783,6 +856,7 @@ export function ChatInterface() {
          }
        }
 
+      flushReasoning(true);
       flushAssistant(true);
       if (!streamedText && (hasFinalPayload || hasDonePayload)) {
         setMessagesForSession(targetSessionKey, (prev) =>
@@ -1046,6 +1120,14 @@ export function ChatInterface() {
                 const isMessageGenerating = isLoading && msgIdx === messages.length - 1;
                 const { markdown, reportHtml } = splitReportHtml(msg.content);
                 const externalReportUrl = extractExternalReport(msg.content);
+                const fallbackThinkingLines = Array.from(new Set(
+                  (msg.progressLogs || []).filter((log) =>
+                    log &&
+                    log !== t('requestSubmittedRouting') &&
+                    log !== t('answerGenerationCompleted')
+                  )
+                ));
+                const displayedThinkingContent = (msg.reasoningContent || "").trim() || fallbackThinkingLines.join("\n");
                 return (
                 <div
                   key={msg.id}
@@ -1065,13 +1147,53 @@ export function ChatInterface() {
                   >
                     {msg.role === "assistant" ? (
                       <>
-                        {msg.reasoningContent && (
+                        {displayedThinkingContent && (
-                          <div className="mb-3 rounded-xl border border-zinc-200 bg-zinc-50/50 p-3 text-sm text-zinc-600 font-mono whitespace-pre-wrap leading-relaxed shadow-inner max-h-[300px] overflow-y-auto">
+                          <div className="mb-3 rounded-xl border border-zinc-200 bg-zinc-50/50 p-3 text-sm text-zinc-600 font-mono whitespace-pre-wrap leading-relaxed shadow-inner">
-                            <div className="flex items-center gap-2 mb-2 text-xs font-semibold text-zinc-500 uppercase tracking-wider">
+                            <button
-                              <Settings className={`h-3.5 w-3.5 ${msg.awaitingFirstToken ? 'animate-spin' : ''}`} />
+                              type="button"
-                              {t('thinkingProcess')}
+                              onClick={() => toggleThinkingCollapsed(msg.id)}
-                            </div>
+                              className="w-full flex items-center justify-between gap-2 mb-2 text-xs font-semibold text-zinc-500 uppercase tracking-wider hover:text-zinc-700 transition-colors"
-                            {msg.reasoningContent}
+                            >
+                              <span className="flex items-center gap-2">
+                                <Settings className={`h-3.5 w-3.5 ${msg.awaitingFirstToken ? 'animate-spin' : ''}`} />
+                                {t('thinkingProcess')}
+                              </span>
+                              <span className="flex items-center gap-2 normal-case text-[11px]">
+                                {msg.usage?.total_tokens ? (
+                                  <span>{t('thinkingTokens', { count: msg.usage.total_tokens })}</span>
+                                ) : msg.reasoningContent ? (
+                                  <span>{t('thinkingCharCount', { count: msg.reasoningContent.length })}</span>
+                                ) : null}
+                                <span
+                                  role="button"
+                                  tabIndex={0}
+                                  onClick={(e) => {
+                                    e.stopPropagation();
+                                    void copyThinkingContent(msg.id, displayedThinkingContent);
+                                  }}
+                                  onKeyDown={(e) => {
+                                    if (e.key === "Enter" || e.key === " ") {
+                                      e.preventDefault();
+                                      e.stopPropagation();
+                                      void copyThinkingContent(msg.id, displayedThinkingContent);
+                                    }
+                                  }}
+                                  className="inline-flex items-center gap-1 rounded-md px-1.5 py-0.5 hover:bg-zinc-200/70 transition-colors"
+                                >
+                                  {thinkingCopiedByMessage[msg.id] ? <Check className="h-3 w-3" /> : <Copy className="h-3 w-3" />}
+                                  <span>{thinkingCopiedByMessage[msg.id] ? t('copied') : t('copy')}</span>
+                                </span>
+                                <span className="inline-flex items-center gap-1">
+                                  {isThinkingCollapsed(msg.id) ? t('expandThinking') : t('collapseThinking')}
+                                  <ChevronDown className={cn("h-3.5 w-3.5 transition-transform", isThinkingCollapsed(msg.id) ? "-rotate-90" : "rotate-0")} />
+                                </span>
+                              </span>
+                            </button>
+                            {!isThinkingCollapsed(msg.id) && (
+                              <div className="max-h-[280px] overflow-y-auto pr-1">
+                                {displayedThinkingContent}
+                              </div>
+                            )}
                           </div>
                         )}
                         {msg.progressLogs && msg.progressLogs.length > 0 ? (

frontend/src/components/Sidebar.tsx

@@ -140,10 +140,8 @@ function Section({
   onRename,
   onTogglePinned,
   onToggleArchived,
-  onBatchDelete,
   activeKey,
   isSelectionMode,
-  setIsSelectionMode,
   selectedKeys,
   setSelectedKeys
 }: {
@@ -153,10 +151,8 @@ function Section({
   onRename: (key: string, currentTitle: string) => void;
   onTogglePinned: (key: string, pinned: boolean) => void;
   onToggleArchived: (key: string, archived: boolean) => void;
-  onBatchDelete: (keys: string[]) => void;
   activeKey: string | null;
   isSelectionMode: boolean;
-  setIsSelectionMode: (val: boolean) => void;
   selectedKeys: string[];
   setSelectedKeys: (val: string[] | ((prev: string[]) => string[])) => void;
 }) {
@@ -684,13 +680,11 @@ function SidebarBody() {
               items={activeSessions} 
               onSelect={handleSelectSession}
               onDelete={handleDeleteSession}
-              onBatchDelete={handleBatchDelete}
               onRename={openRenameDialog}
               onTogglePinned={handleTogglePinned}
               onToggleArchived={handleToggleArchived}
               activeKey={activeSessionKey}
               isSelectionMode={activeSelectionMode}
-              setIsSelectionMode={setActiveSelectionMode}
               selectedKeys={activeSelectedKeys}
               setSelectedKeys={setActiveSelectedKeys}
             />
@@ -714,13 +708,11 @@ function SidebarBody() {
                 items={archivedSessions} 
                 onSelect={handleSelectSession}
                 onDelete={handleDeleteSession}
-                onBatchDelete={handleBatchDelete}
                 onRename={openRenameDialog}
                 onTogglePinned={handleTogglePinned}
                 onToggleArchived={handleToggleArchived}
                 activeKey={activeSessionKey}
                 isSelectionMode={archivedSelectionMode}
-                setIsSelectionMode={setArchivedSelectionMode}
                 selectedKeys={archivedSelectedKeys}
                 setSelectedKeys={setArchivedSelectedKeys}
               />

frontend/src/i18n/locales/en.json

@@ -36,6 +36,10 @@
   "processing": "Processing...",
   "processCompleted": "Completed",
   "thinkingProcess": "Thinking Process",
+  "thinkingTokens": "{{count}} tokens",
+  "thinkingCharCount": "{{count}} chars",
+  "expandThinking": "Expand",
+  "collapseThinking": "Collapse",
   "modelThinking": "Model is thinking, please wait...",
   "openReportInNewTab": "Open report in new tab",
   "artifactPreview": "File Preview",

frontend/src/i18n/locales/zh.json

@@ -49,6 +49,10 @@
   "processing": "正在处理中",
   "processCompleted": "处理完成",
   "thinkingProcess": "思考过程",
+  "thinkingTokens": "{{count}} tokens",
+  "thinkingCharCount": "{{count}} 字",
+  "expandThinking": "展开",
+  "collapseThinking": "收起",
   "modelThinking": "模型思考中，请稍候...",
   "openReportInNewTab": "在新标签页中打开分析报告",
   "artifactPreview": "文件预览",

frontend/src/pages/Dashboard.tsx

@@ -9,7 +9,7 @@ import { ScrollArea } from "@/components/ui/scroll-area";
 import { Table, TableBody, TableCell, TableHead, TableHeader, TableRow } from "@/components/ui/table";
 import { Popover, PopoverContent, PopoverTrigger } from "@/components/ui/popover";
 import { Input } from "@/components/ui/input";
-import { X, Type, AlignLeft, AlignCenter, AlignRight, Bold, Italic, Underline } from "lucide-react";
+import { X, Type, Bold, Italic, Underline } from "lucide-react";
 import { BarChart, Bar, XAxis, YAxis, CartesianGrid, Tooltip, ResponsiveContainer, LineChart, Line } from 'recharts';
 import { VegaChart } from "@/components/VegaChart";
 import 'react-grid-layout/css/styles.css';

frontend/src/pages/Login.tsx

@@ -87,10 +87,8 @@ export function Login() {
           </h1>
           <div className="absolute right-0 bottom-0 translate-y-4">
             <DropdownMenu>
-              <DropdownMenuTrigger asChild>
+              <DropdownMenuTrigger className="h-9 w-9 rounded-full bg-white/50 backdrop-blur-sm shadow-sm border border-zinc-200/50 text-zinc-500 hover:text-zinc-900 hover:bg-white transition-all inline-flex items-center justify-center">
-                <Button variant="ghost" size="icon" className="h-9 w-9 rounded-full bg-white/50 backdrop-blur-sm shadow-sm border border-zinc-200/50 text-zinc-500 hover:text-zinc-900 hover:bg-white transition-all">
+                <Languages className="h-4 w-4" />
-                  <Languages className="h-4 w-4" />
-                </Button>
               </DropdownMenuTrigger>
               <DropdownMenuContent align="end" className="w-32">
                 <DropdownMenuItem onClick={() => i18n.changeLanguage('zh')} className={i18n.language === 'zh' ? 'bg-zinc-100' : ''}>