v2.8.0 清空历史真正物理删除 + 请求超时可配修复慢接口被掐断

1. 清空历史=真正清空（不再只是删行留痕）:
   - 清空前先停止所有运行中测试线程，避免其继续写入孤儿数据
   - DELETE 后重置自增序列(sqlite_sequence)，新测试 ID 从 1 开始
   - WAL checkpoint(TRUNCATE) + VACUUM 物理回收文件空间，数据不可恢复
   - 单条删除删光时同样重置自增序列

2. 慢接口被掐断修复（预处理/首字耗时过长 Read timed out）:
   - 读取超时默认 300s→1800s(30分钟)，连接 60s
   - 「速度测试配置」新增 请求超时(秒) 输入框(默认1800)，测试连接+速度采样共用，按次可调(如3600)
   - 超时错误改为友好提示：等待响应超时(超过N秒未收到数据)，引导调大请求超时，不再裸抛 Read timed out
   - 测试启动日志打印当前超时配置
   - 修复案例: 18008 Qwen3.8-FP8 131072 长上下文预填充被 300s 掐断

API.md

@@ -137,20 +137,35 @@
 
 **响应：** `{ "ok": true, "id": 9 }`
 
-### `GET /api/tests?limit=<n>`
-测试历史列表（按 id 倒序）。`limit` 默认 100，最大 1000。
+### `GET /api/tests`
+测试历史列表，支持**分页 + 搜索 + 筛选 + 排序**：
+
+| 参数 | 说明 |
+|------|------|
+| `page` | 页码，默认 1 |
+| `page_size` | 每页条数，默认 20，最大 100 |
+| `q` | 关键词搜索（匹配 id/名称/模型/提供商/状态/时间） |
+| `status` | 状态筛选，逗号多选：`done,error,canceled,running` |
+| `provider` | 提供商筛选：`openai` / `anthropic` / `google` |
+| `sort` | 排序字段：`id`/`created_at`/`name`/`provider`/`model`/`status`/`samples`/`ttft`/`prefill`/`decode`/`total_ms`，默认 `id` |
+| `order` | 排序方向：`asc` / `desc`，默认 `desc`（数值字段空值恒排最后） |
 
 **响应：**
 ```json
-[
-  { "id": 9, "created_at": "2026-08-23 18:52:00", "status": "done",
+{ "ok": true, "items": [ { "id": 9, "created_at": "2026-08-23 18:52:00", "status": "done",
     "provider": "openai", "model": "unsloth/Qwen3.8-27B-Q4_K_M",
     "name": "Qwen3 不同上下文长度速度对比",
     "error": "",
-    "summary": { "samples_ok": 2, "samples_total": 2, "avg_ttft_ms": 1808.7, ... } }
-]
+    "summary": { "samples_ok": 2, "samples_total": 2, "avg_ttft_ms": 1808.7, ... } } ],
+  "total": 31, "page": 1, "page_size": 20, "total_pages": 2 }
 ```
 
+### `DELETE /api/tests`
+一键清空全部测试历史（测试记录 + 采样指标 + 日志），不可恢复。响应 `{ "ok": true }`。
+
+### `GET /api/tests/export`
+一键导出全部历史记录为 JSON 文件下载（`Content-Disposition: attachment`，文件名带时间戳）。每条含基本字段 + 配置（**API Key 打码为 `******`**）+ 生成参数 + 汇总指标，不含逐条采样与完整日志。
+
 ### `POST /api/tests/<id>/cancel`
 停止正在运行的测试。响应 `{ "ok": true, "msg": "正在停止..." }`
 
@@ -353,7 +368,12 @@ curl -X POST $BASE/api/tests -H 'Content-Type: application/json' -d '{
 }'
 
 # 查询测试列表 / 详情
-curl "$BASE/api/tests?limit=10"
+curl "$BASE/api/tests?page=1&page_size=20"
+curl "$BASE/api/tests?q=Qwen&status=done,error&sort=decode&order=desc&page=1&page_size=10"
+# 一键清空全部历史（不可恢复）
+curl -X DELETE $BASE/api/tests
+# 一键导出全部历史（JSON 文件下载）
+curl -OJ $BASE/api/tests/export
 curl $BASE/api/tests/9
 
 # 画图数据（CSV）

README.md

@@ -4,7 +4,7 @@
 
 - **访问地址：** `http://<IP>:16097/`
 - **技术栈：** Python 3 + Flask + SQLite（纯 REST，无额外依赖）
-- **版本：** v2.5.0
+- **版本：** v2.7.0
 
 ---
 
@@ -30,7 +30,8 @@
 ### 📊 指标与结果
 - 实时指标卡：首字延迟、预填充速度、解码速度、上文/输出 tokens、总耗时
 - **整体统计（平均/最小/最大）**：详情弹窗与 Excel 汇总展示首字延迟、预填充速度、解码速度、总耗时的平均/最小/最大
-- **推理型模型兼容**：支持 Qwen3 / DeepSeek 等思维链模型（`reasoning_content` / `thinking` / `thought`），思维过程计入输出，不会误报“未收到输出”
+- **推理型模型兼容**：支持 Qwen3 / DeepSeek 等思维链模型（`reasoning_content` / `reasoning` / `thinking` / `thought`），思维过程计入输出，不会误报“未收到输出”
+- **解码速度防荒谬值**：兼容 vLLM/Qwen3.8-27B 用 `reasoning` 字段流式输出思维链（此前未识别导致 output_chars=0、解码时间≈0、解码速度飙到数百万 tok/s，已修复）；另防“幻影 token”——usage 报了输出 token 但流里没收到任何文本时（如推理用满 max_tokens 没正文），解码速度标记为不可测（—）而非荒谬数字
 - **采样失败不中断**：单次采样失败会记录并继续，不会让整个测试半途终止；全部失败才标记 error
 - 实时控制台日志：校准、预热、每次采样明细全程可追溯
 - **每次完整测试**支持：
@@ -39,6 +40,11 @@
   - **文件下载 Excel（xlsx）**：历史记录「Excel」按钮或详情弹窗「导出 Excel」，包含 汇总 / 采样明细 / 日志 三个 Sheet
   - **文件下载 JSON**：详情弹窗「导出 JSON」
 - 测试历史留存（含测试名称），可随时刷新、查看、导出、删除
+- **📥 一键导出全部历史**：历史卡片「导出全部」按钮，把全部测试记录（配置 Key 打码 + 生成参数 + 汇总指标）导出为一个 JSON 文件下载（`GET /api/tests/export`）
+- **📄 历史分页 + 筛选 + 搜索 + 排序**：历史列表支持分页（10/20/50/100 条每页）、多条件筛选（状态多选、提供商）、关键词搜索（编号/名称/模型/提供商/状态/时间）、点击表头升降排序（#/时间/名称/提供商/模型/采样/首字/预填充/解码/状态，数值列空值恒排最后）；提供「✕ 重置」一键恢复默认
+- **🗑 一键清空历史**：历史卡片右上角「清空历史」二次确认后，删除全部测试记录/采样/日志（不可恢复）
+- **🗜 左侧配置块可折叠**：大模型接口配置 / 速度测试配置 / 评测站同步 三个卡片点击标题即可折叠/展开（箭头指示），状态本地持久化
+- **⬆⬇ 一键置顶/置底**：右下角固定悬浮按钮（亮蓝渐变高对比配色，易于发现），平滑滚动到页面顶部/底部
 - **⚖️ 多测试结果对比**：测试历史表格勾选多个测试（可全选），点「⚖️ 对比所选」弹出对比面板：
   - **指标对比表**：时间 / 名称 / 模型 / 并发档 / 采样 / 首字 / 预填充 / 解码 / 单流均解码 / 输出 / 总耗时 并排展示（点击测试名可跳详情）
   - **柱状图**：各测试预填充（空心柱）vs 解码（实心柱）速度对比
@@ -100,7 +106,9 @@ pip install -r requirements.txt
 | GET/POST | `/api/configs` | 配置列表 / 新增配置 |
 | GET/PUT/DELETE | `/api/configs/<id>` | 单个配置 / 更新（局部）/ 删除 |
 | POST | `/api/configs/test` | 测试连接 |
-| GET/POST | `/api/tests?limit=n` | 测试历史 / 启动测试（异步） |
+| GET/POST | `/api/tests?page=&page_size=&q=&status=&provider=&sort=&order=` | 测试历史（分页/搜索/筛选/排序）/ 启动测试（异步） |
+| DELETE | `/api/tests` | 一键清空全部测试历史（含采样与日志） |
+| GET | `/api/tests/export` | 一键导出全部历史记录（JSON 文件下载，API Key 打码） |
 | GET | `/api/tests/<id>` | 测试详情（含 runs / logs / summary / 按长度分组） |
 | GET | `/api/tests/<id>/logs?after=<id>` | 增量日志（前端轮询用） |
 | POST | `/api/tests/<id>/cancel` | 停止测试 |
@@ -178,4 +186,4 @@ llm-speed-tester/
 ## Git
 
 - **仓库：** `hz4th_coder/llm-speed-tester`
-- **版本：** v2.5.0（模型实时获取：接口配置区「📋 查看模型」按钮，Base URL+API Key 下实时拉取模型列表（GET /models），搜索/点击填入，仍支持手动输入）
+- **版本：** v2.7.0（一键导出全部历史(JSON)；左侧三块配置卡片可折叠展开(状态持久化)；置顶/置底按钮高亮蓝渐变配色）

app.py

@@ -3,6 +3,8 @@
 import io
 import json
 import subprocess
+import time
+from datetime import datetime
 
 import requests
 from flask import Flask, jsonify, request, send_file, send_from_directory
@@ -111,13 +113,17 @@ def list_models_api():
 
 @app.route("/api/configs/test", methods=["POST"])
 def test_config():
-    cfg = _fill_defaults(request.get_json(force=True) or {})
+    body = request.get_json(force=True) or {}
+    cfg = _fill_defaults(body)
     if not cfg.get("api_key"):
         return jsonify({"ok": False, "error": "请填写 API Key"}), 400
     try:
         # 连接测试：只要流式请求成功返回（哪怕正文为空/只有思维链）都算连通
+        # 慢接口长预处理可能长时间等不到首字，这里直接用请求里带的超时（默认 30 分钟），避免误判接口不通
         m = call_stream(cfg, "你好，请简要回答：1+1=？",
-                        {"max_tokens": 32, "avoid_cache": False})
+                        {"max_tokens": 32, "avoid_cache": False,
+                         "read_timeout": body.get("read_timeout"),
+                         "connect_timeout": body.get("connect_timeout")})
         note = ""
         if not (m.get("output_tokens") or m.get("output_chars")):
             note = "（连接正常，但本次未返回正文内容，可能为推理型模型）"
@@ -146,13 +152,63 @@ def start_test():
     return jsonify({"ok": True, "id": tid})
 
 
+@app.route("/api/tests/export")
+def export_tests():
+    """一键导出全部历史记录（JSON 文件下载，API Key 打码）"""
+    data = db.export_tests()
+    payload = {
+        "ok": True,
+        "count": len(data),
+        "exported_at": datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
+        "tests": data,
+    }
+    resp = app.response_class(json.dumps(payload, ensure_ascii=False),
+                              mimetype="application/json")
+    resp.headers["Content-Disposition"] = (
+        'attachment; filename="llm_speed_tests_%s.json"'
+        % datetime.now().strftime("%Y%m%d_%H%M%S"))
+    return resp
+
+
 @app.route("/api/tests", methods=["GET"])
 def list_tests():
+    """历史记录列表：支持关键词搜索 q、状态 status(逗号多选)、提供商 provider、
+    字段排序 sort/order、分页 page/page_size"""
     try:
-        limit = int(request.args.get("limit", 100))
+        page = max(1, int(request.args.get("page", 1)))
+        page_size = min(100, max(1, int(request.args.get("page_size", 20))))
     except ValueError:
-        limit = 100
-    return jsonify(db.list_tests(max(1, min(limit, 1000))))
+        page, page_size = 1, 20
+    q = (request.args.get("q") or "").strip()
+    status = (request.args.get("status") or "").strip()
+    provider = (request.args.get("provider") or "").strip()
+    sort = (request.args.get("sort") or "id").strip()
+    order = (request.args.get("order") or "desc").strip().lower()
+    if order not in ("asc", "desc"):
+        order = "desc"
+    data = db.query_tests(q=q, status=status, provider=provider,
+                          sort=sort, order=order, page=page, page_size=page_size)
+    return jsonify({"ok": True, **data})
+
+
+@app.route("/api/tests", methods=["DELETE"])
+def clear_tests():
+    """一键清空全部测试历史（含采样指标与日志）。
+    先停止所有运行中的测试线程，避免清空后它们继续写入孤儿数据；
+    再由 db.clear_tests 做物理删除（重置自增 + VACUUM 瘦身），真正清空、不可恢复。"""
+    # 1) 停止所有运行中的测试，并等待其退出
+    alive = [r for r in RUNNERS.values() if r.is_alive()]
+    for r in alive:
+        r.request_cancel()
+    if alive:
+        deadline = time.time() + 5
+        for r in alive:
+            if r.is_alive():
+                r.join(timeout=max(0.1, deadline - time.time()))
+    RUNNERS.clear()
+    # 2) 物理清空数据库
+    db.clear_tests()
+    return jsonify({"ok": True})
 
 
 @app.route("/api/tests/<int:tid>", methods=["GET"])

config.py

@@ -10,9 +10,11 @@ DATA_DIR = os.path.join(BASE_DIR, "data")
 LOG_DIR = os.path.join(BASE_DIR, "logs")
 DB_PATH = os.path.join(DATA_DIR, "llm_speed_tester.db")
 
-# 流式请求超时：连接 60s，两次数据包间隔最长 300s（推理型/长上文模型也够用）
+# 流式请求超时：连接 60s，两次数据包间隔最长 1800s（30 分钟）
+# 注意：慢接口长上下文预处理（预填充）期间服务端可能长时间不返回任何字节，
+# 等待首字受 STREAM_READ_TIMEOUT 限制；如接口更慢可在「速度测试配置」里按次调大请求超时。
 CONNECT_TIMEOUT = 60
-STREAM_READ_TIMEOUT = 300
+STREAM_READ_TIMEOUT = 1800
 
 # data-chart-tool 图表服务地址（用它的 /api/chart 画折线图：预填充左轴虚线 / 解码右轴实线）
 CHART_API_BASE = "http://127.0.0.1:16016"

database.py

@@ -2,6 +2,7 @@
 """SQLite 存储：提供商配置 / 测试记录 / 每次采样指标 / 全量日志"""
 import os
 import json
+import math
 import sqlite3
 import threading
 import time
@@ -204,20 +205,93 @@ def get_test(tid: int):
             conn.close()
 
 
-def list_tests(limit=100):
+# 历史列表排序字段 → 取值函数（数值字段取 summary 聚合值，None 恒排最后）
+SORT_KEYS = {
+    "id": lambda t: t["id"],
+    "created_at": lambda t: t["created_at"],
+    "name": lambda t: (t.get("name") or "").lower(),
+    "provider": lambda t: (t.get("provider") or "").lower(),
+    "model": lambda t: (t.get("model") or "").lower(),
+    "status": lambda t: t.get("status") or "",
+    "samples": lambda t: (t.get("summary") or {}).get("samples_ok"),
+    "ttft": lambda t: (t.get("summary") or {}).get("avg_ttft_ms"),
+    "prefill": lambda t: (t.get("summary") or {}).get("avg_prefill_speed"),
+    "decode": lambda t: (t.get("summary") or {}).get("avg_decode_speed"),
+    "total_ms": lambda t: (t.get("summary") or {}).get("avg_total_ms"),
+}
+
+
+def query_tests(q="", status="", provider="", sort="id", order="desc",
+                page=1, page_size=20):
+    """历史记录：关键词搜索 + 状态/提供商筛选 + 字段排序 + 分页。
+    返回 {"items": [...], "total": N, "page": p, "page_size": s, "total_pages": n}"""
     with _lock:
         conn = _connect()
         try:
+            where, args = [], []
+            if q:
+                like = "%" + q + "%"
+                where.append("(id LIKE ? OR name LIKE ? OR model LIKE ? OR provider LIKE ?"
+                             " OR status LIKE ? OR created_at LIKE ?)")
+                args += [like] * 6
+            if status:
+                sts = [s.strip() for s in status.split(",") if s.strip()]
+                if sts:
+                    where.append("status IN (%s)" % ",".join("?" * len(sts)))
+                    args += sts
+            if provider:
+                where.append("provider = ?")
+                args.append(provider)
+            wsql = ("WHERE " + " AND ".join(where)) if where else ""
+
             rows = conn.execute(
                 "SELECT id,created_at,status,provider,model,name,summary_json,gen_cfg_json,error "
-                "FROM tests ORDER BY id DESC LIMIT ?", (limit,)).fetchall()
+                "FROM tests " + wsql + " ORDER BY id DESC", args).fetchall()
             out = []
             for r in rows:
                 d = dict(r)
                 d["summary"] = json.loads(d.pop("summary_json") or "{}")
                 d["gen"] = json.loads(d.pop("gen_cfg_json") or "{}")
                 out.append(d)
-            return out
+
+            # Python 侧排序：None 值恒排最后（无论升降序）
+            keyfn = SORT_KEYS.get(sort) or SORT_KEYS["id"]
+            present = [t for t in out if keyfn(t) is not None]
+            absent = [t for t in out if keyfn(t) is None]
+            present.sort(key=keyfn, reverse=(order == "desc"))
+            out = present + absent
+
+            total = len(out)
+            total_pages = max(1, math.ceil(total / page_size))
+            page = min(max(1, page), total_pages)
+            start = (page - 1) * page_size
+            return {
+                "items": out[start:start + page_size],
+                "total": total,
+                "page": page,
+                "page_size": page_size,
+                "total_pages": total_pages,
+            }
+        finally:
+            conn.close()
+
+
+def clear_tests():
+    """一键清空全部历史（测试记录 + 采样指标 + 日志）。
+    真正清空：删除全部行 + 重置自增序列 + WAL 检查点 + VACUUM 物理回收文件空间，
+    数据不可恢复（不是只从列表里移除，文件里也不再残留）。"""
+    with _lock:
+        conn = _connect()
+        try:
+            conn.execute("DELETE FROM test_runs")
+            conn.execute("DELETE FROM logs")
+            conn.execute("DELETE FROM tests")
+            # 重置自增序列：清空后新测试 ID 从 1 重新开始
+            conn.execute("DELETE FROM sqlite_sequence WHERE name IN ('tests','test_runs','logs')")
+            conn.commit()
+            # WAL 检查点截断 + VACUUM：把文件里已删除的数据页物理抹掉、文件瘦身
+            conn.execute("PRAGMA wal_checkpoint(TRUNCATE)")
+            conn.execute("VACUUM")
         finally:
             conn.close()
 
@@ -229,11 +303,39 @@ def delete_test(tid: int):
             conn.execute("DELETE FROM tests WHERE id=?", (tid,))
             conn.execute("DELETE FROM test_runs WHERE test_id=?", (tid,))
             conn.execute("DELETE FROM logs WHERE test_id=?", (tid,))
+            # 全部删光时顺带重置自增序列
+            n = conn.execute("SELECT COUNT(*) FROM tests").fetchone()[0]
+            if n == 0:
+                conn.execute("DELETE FROM sqlite_sequence WHERE name IN ('tests','test_runs','logs')")
             conn.commit()
         finally:
             conn.close()
 
 
+def export_tests():
+    """导出全部历史：每条含 基本字段 + 配置(Key打码) + 生成参数 + 摘要指标，
+    不含逐条采样与完整日志（避免文件过大）"""
+    with _lock:
+        conn = _connect()
+        try:
+            rows = conn.execute(
+                "SELECT id,created_at,status,provider,model,name,config_json,gen_cfg_json,summary_json,error "
+                "FROM tests ORDER BY id").fetchall()
+            out = []
+            for r in rows:
+                d = dict(r)
+                cfg = json.loads(d.pop("config_json") or "{}")
+                if cfg.get("api_key"):
+                    cfg["api_key"] = "******"
+                d["config"] = cfg
+                d["gen"] = json.loads(d.pop("gen_cfg_json") or "{}")
+                d["summary"] = json.loads(d.pop("summary_json") or "{}")
+                out.append(d)
+            return out
+        finally:
+            conn.close()
+
+
 # ───────────────────────── 采样指标 ─────────────────────────
 
 def add_run(tid: int, run_index: int, metrics: dict, error: str = "", context_length: int = 0):

llm_providers.py

@@ -63,6 +63,10 @@ def _metrics(start, first_token_at, end, prompt_tokens, output_tokens,
     total_ms = (end - start) * 1000
     prefill = (prompt_tokens / (ttft_ms / 1000)) if prompt_tokens and ttft_ms > 0 else None
     decode = (output_tokens / (decode_ms / 1000)) if output_tokens and decode_ms > 0 else None
+    # 防“幻影 token”：usage 报了输出 token，但整个流没收到任何文本（如推理全用满 max_tokens
+    # 或提供商只回 usage 不正文），此时解码时间无意义，不报出数百万的荒谬速度
+    if output_tokens and decode is not None and not output_chars:
+        decode = None
     return {
         "prompt_tokens": int(prompt_tokens or 0),
         "output_tokens": int(output_tokens or 0),
@@ -112,7 +116,7 @@ def stream_openai(cfg, prompt, gen, log, should_stop=None):
             if should_stop and should_stop():
                 raise StopRequested()
             resp = requests.post(url, json=build_payload(), headers=headers, stream=True,
-                                 timeout=(config.CONNECT_TIMEOUT, config.STREAM_READ_TIMEOUT))
+                                 timeout=_timeouts(gen))
             if resp.status_code == 200:
                 break
             err = resp.text[:400]
@@ -132,8 +136,9 @@ def stream_openai(cfg, prompt, gen, log, should_stop=None):
             event_count += 1
             if obj.get("choices"):
                 delta = obj["choices"][0].get("delta") or {}
-                # 兼容推理型模型：Qwen3/DeepSeek 思维链在 reasoning_content
-                piece = delta.get("content") or delta.get("reasoning_content") or ""
+                # 兼容推理型模型思维链：DeepSeek 用 reasoning_content，部分 vLLM(Qwen3.8-27B-FP8) 用 reasoning，Anthropic/Gemini 分别在各自适配器处理
+                piece = (delta.get("content") or delta.get("reasoning_content")
+                         or delta.get("reasoning") or "")
                 if piece:
                     if first_token_at is None:
                         first_token_at = time.time()
@@ -147,6 +152,10 @@ def stream_openai(cfg, prompt, gen, log, should_stop=None):
                     cached_tokens = details.get("cached_tokens") or 0
     except StopRequested:
         raise
+    except requests.exceptions.ReadTimeout:
+        raise ProviderError(_read_timeout_err(gen))
+    except requests.exceptions.ConnectTimeout:
+        raise ProviderError(_connect_timeout_err(gen))
     except Exception as e:
         raise ProviderError("流式请求异常: %s" % e)
     finally:
@@ -166,6 +175,30 @@ def _bad_stream_options(err: str):
             or "additional properties" in err)
 
 
+def _timeouts(gen):
+    """从测试参数 gen 里取超时配置（秒），未配置则用全局默认。
+    返回 (connect, read) 元组，供 requests timeout 使用。"""
+    try:
+        connect = float(gen.get("connect_timeout") or config.CONNECT_TIMEOUT)
+    except (TypeError, ValueError):
+        connect = config.CONNECT_TIMEOUT
+    try:
+        read = float(gen.get("read_timeout") or config.STREAM_READ_TIMEOUT)
+    except (TypeError, ValueError):
+        read = config.STREAM_READ_TIMEOUT
+    return (max(connect, 5), max(read, 10))
+
+
+def _read_timeout_err(gen):
+    """读超时的友好报错：提示这是等待数据超时，可调大请求超时"""
+    return ("等待响应超时：超过 %.0f 秒未收到数据（接口预处理/首字耗时过长或网络过慢），"
+            "请在「速度测试配置」中调大“请求超时（秒）”" % _timeouts(gen)[1])
+
+
+def _connect_timeout_err(gen):
+    return "连接超时：超过 %.0f 秒未建立连接（请检查地址/网络）" % _timeouts(gen)[0]
+
+
 def _iter_json(resp):
     """解析 SSE data: 行，逐个返回 JSON 对象"""
     for raw in resp.iter_lines(decode_unicode=True):
@@ -201,7 +234,7 @@ def stream_anthropic(cfg, prompt, gen, log, should_stop=None):
         if should_stop and should_stop():
             raise StopRequested()
         resp = requests.post(url, json=payload, headers=headers, stream=True,
-                             timeout=(config.CONNECT_TIMEOUT, config.STREAM_READ_TIMEOUT))
+                             timeout=_timeouts(gen))
         if resp.status_code != 200:
             err = resp.text[:400]
             resp.close()
@@ -228,6 +261,10 @@ def stream_anthropic(cfg, prompt, gen, log, should_stop=None):
                 output_tokens = usage.get("output_tokens") or output_tokens
     except StopRequested:
         raise
+    except requests.exceptions.ReadTimeout:
+        raise ProviderError(_read_timeout_err(gen))
+    except requests.exceptions.ConnectTimeout:
+        raise ProviderError(_connect_timeout_err(gen))
     except Exception as e:
         raise ProviderError("流式请求异常: %s" % e)
     finally:
@@ -267,7 +304,7 @@ def stream_google(cfg, prompt, gen, log, should_stop=None):
         if should_stop and should_stop():
             raise StopRequested()
         resp = requests.post(url, params=params, json=payload, headers=headers, stream=True,
-                             timeout=(config.CONNECT_TIMEOUT, config.STREAM_READ_TIMEOUT))
+                             timeout=_timeouts(gen))
         if resp.status_code != 200:
             err = resp.text[:400]
             resp.close()
@@ -294,6 +331,10 @@ def stream_google(cfg, prompt, gen, log, should_stop=None):
                 cached_tokens = um.get("cachedContentTokenCount") or 0
     except StopRequested:
         raise
+    except requests.exceptions.ReadTimeout:
+        raise ProviderError(_read_timeout_err(gen))
+    except requests.exceptions.ConnectTimeout:
+        raise ProviderError(_connect_timeout_err(gen))
     except Exception as e:
         raise ProviderError("流式请求异常: %s" % e)
     finally:

static/css/style.css

@@ -43,17 +43,44 @@ body {
 
 /* ── 布局 ── */
 .layout { display: grid; grid-template-columns: 340px 1fr; gap: 16px; padding: 16px 24px; align-items: start; }
-.config-panel { display: flex; flex-direction: column; gap: 16px; position: sticky; top: 72px; }
+.config-panel {
+  display: flex; flex-direction: column; gap: 16px;
+  position: sticky; top: 72px;
+  /* 左侧独立滚动：面板固定视口高度，鼠标在左侧滚动时只翻左侧几个块，不带动右侧 */
+  max-height: calc(100vh - 88px);
+  overflow-y: auto;
+  overscroll-behavior: contain;
+  padding-right: 6px;
+  scrollbar-width: thin;
+  scrollbar-color: var(--border) transparent;
+}
+.config-panel::-webkit-scrollbar { width: 8px; }
+.config-panel::-webkit-scrollbar-thumb { background: var(--border); border-radius: 4px; }
+.config-panel::-webkit-scrollbar-thumb:hover { background: var(--accent); }
 .main { display: flex; flex-direction: column; gap: 16px; min-width: 0; }
 
 .card {
   background: var(--panel); border: 1px solid var(--border);
   border-radius: 12px; padding: 16px;
 }
+/* ── 可折叠卡片（左侧三块） ── */
 .card h2 { font-size: 15px; margin-bottom: 14px; }
+.collapse-head {
+  display: flex; align-items: center; justify-content: space-between;
+  cursor: pointer; user-select: none;
+}
+.collapse-head:hover { color: var(--accent); }
+.collapse-arrow {
+  font-size: 12px; color: var(--muted); transition: transform .15s;
+}
+.card.collapsed .collapse-arrow { transform: rotate(-90deg); }
+.card.collapsed .collapse-body { display: none; }
+
+/* ── 卡片头部 ── */
 .card-head { display: flex; align-items: center; justify-content: space-between; margin-bottom: 14px; }
 .card-head h2 { margin-bottom: 0; }
 
+
 /* ── 表单 ── */
 .field { margin-bottom: 12px; }
 .field label { display: block; font-size: 12px; color: var(--muted); margin-bottom: 5px; }
@@ -184,6 +211,17 @@ body {
 .console .ln.sys { color: #5b6b8c; font-style: italic; }
 
 /* ── 历史表 ── */
+.hist-filters {
+  display: flex; flex-wrap: wrap; gap: 6px; align-items: center;
+  margin-bottom: 10px;
+}
+.hist-filters select {
+  background: var(--panel2); color: var(--text); border: 1px solid var(--border);
+  border-radius: 8px; padding: 7px 10px; font-size: 12.5px; outline: none;
+}
+.hist-filters select:focus { border-color: var(--accent); }
+.hist-filters .icon-input { min-width: 160px; }
+
 .table-wrap { overflow-x: auto; }
 table.history { width: 100%; border-collapse: collapse; font-size: 13px; }
 .history th, .history td { padding: 8px 10px; text-align: left; border-bottom: 1px solid var(--border); white-space: nowrap; }
@@ -192,6 +230,17 @@ table.history { width: 100%; border-collapse: collapse; font-size: 13px; }
 .history td.num { font-family: var(--mono); }
 .history .chk-col { width: 34px; text-align: center; }
 .history .chk-col input { width: 15px; height: 15px; cursor: pointer; accent-color: var(--accent); }
+.history th.sortable { cursor: pointer; user-select: none; }
+.history th.sortable:hover { color: var(--accent); }
+.history th.sortable.asc::after { content: " ▲"; font-size: 10px; }
+.history th.sortable.desc::after { content: " ▼"; font-size: 10px; }
+
+/* 历史分页条 */
+.hist-pager {
+  display: flex; align-items: center; justify-content: center; gap: 14px;
+  margin-top: 12px; padding-top: 10px; border-top: 1px solid var(--border);
+}
+.hist-pager .btn:disabled { opacity: .4; cursor: not-allowed; }
 .status-pill { padding: 2px 10px; border-radius: 12px; font-size: 11px; }
 .status-pill.running { background: rgba(79,140,255,.15); color: var(--accent); }
 .status-pill.done { background: rgba(34,197,139,.15); color: var(--accent2); }
@@ -257,8 +306,22 @@ pre.json-box {
   padding: 10px; font: 11.5px/1.6 var(--mono); overflow: auto; max-height: 260px; color: #a8f0d0;
 }
 
+/* ── 右下角一键置顶/置底（高对比配色，易于发现） ── */
+.scroll-fab {
+  position: fixed; right: 20px; bottom: 20px;
+  display: flex; flex-direction: column; gap: 8px; z-index: 60;
+}
+.fab {
+  width: 44px; height: 44px; border-radius: 50%; border: none;
+  background: linear-gradient(135deg, #5a92ff, #2f6bff);
+  color: #fff; font-size: 18px; cursor: pointer;
+  box-shadow: 0 6px 18px rgba(47,107,255,.55); transition: .15s; line-height: 1;
+}
+.fab:hover { filter: brightness(1.15); transform: translateY(-1px); }
+.fab:active { transform: translateY(0); }
+
 @media (max-width: 1000px) {
   .layout { grid-template-columns: 1fr; }
-  .config-panel { position: static; }
+  .config-panel { position: static; max-height: none; overflow: visible; }
   .metrics { grid-template-columns: repeat(3, 1fr); }
 }

static/index.html

@@ -19,8 +19,9 @@
   <main class="layout">
     <!-- 左侧：配置 -->
     <aside class="config-panel">
-      <section class="card">
-        <h2>🔌 大模型接口配置</h2>
+      <section class="card collapsible" id="sec-interface">
+        <h2 class="collapse-head">🔌 大模型接口配置<span class="collapse-arrow">▾</span></h2>
+        <div class="collapse-body">
         <div class="field">
           <label>提供商类型</label>
           <select id="cfg-provider">
@@ -75,10 +76,12 @@
         </div>
         <button class="btn block" id="btn-test-conn">🔍 测试连接</button>
         <div class="conn-result" id="conn-result" hidden></div>
+        </div>
       </section>
 
-      <section class="card">
-        <h2>🚀 速度测试配置</h2>
+      <section class="card collapsible" id="sec-test">
+        <h2 class="collapse-head">🚀 速度测试配置<span class="collapse-arrow">▾</span></h2>
+        <div class="collapse-body">
         <div class="field">
           <label>测试名称（主题）</label>
           <input id="gen-name" type="text" placeholder="如：DeepSeek-V4 不同上下文长度速度对比">
@@ -122,13 +125,20 @@
             <span class="slider"></span>
           </label>
         </div>
+        <div class="field">
+          <label>请求超时（秒）<span class="hint-inline">等待首字/预处理上限，接口慢就调大</span></label>
+          <div class="icon-input"><span class="icon">⏱</span><input id="gen-timeout" type="number" min="30" step="30" value="1800"></div>
+          <div class="hint">连接测试与速度采样共用：超过此时长仍未收到数据才判定超时（默认 1800 秒 = 30 分钟）。测慢接口/超长上下文时建议调大，如 3600。</div>
+        </div>
         <div class="btn-group">
           <button class="btn primary block" id="btn-start">▶ 开始测试</button>
           <button class="btn danger block" id="btn-cancel" disabled>■ 停止</button>
         </div>
+        </div>
       </section>
-      <section class="card">
-        <h2>📤 评测站同步</h2>
+      <section class="card collapsible" id="sec-eval">
+        <h2 class="collapse-head">📤 评测站同步<span class="collapse-arrow">▾</span></h2>
+        <div class="collapse-body">
         <div class="field">
           <label>评测站地址</label>
           <div class="icon-input"><span class="icon">🌐</span><input id="eval-url" placeholder="http://127.0.0.1:16066"></div>
@@ -147,6 +157,7 @@
         <button class="btn block" id="btn-eval-test">🔗 测试连接</button>
         <div class="conn-result" id="eval-result" hidden></div>
         <div class="hint" style="margin-top:6px">测试完成后，在历史列表或详情里点「📤 发送到评测站」，一键把结果发布到模型评测网站（端口 16066）对应账号下。</div>
+        </div>
       </section>
     </aside>
 
@@ -175,23 +186,63 @@
 
       <div class="card">
         <div class="card-head">
-          <h2>🗂 测试历史</h2>
+          <h2>🗂 测试历史 <span class="hint" id="hist-total"></span></h2>
           <div class="log-actions">
             <span class="hint" id="hist-selected"></span>
             <button class="btn small primary" id="btn-compare">⚖️ 对比所选</button>
+            <button class="btn small" id="btn-export-history" title="一键导出全部历史记录（JSON）">📥 导出全部</button>
+            <button class="btn small danger" id="btn-clear-history" title="一键清空全部历史记录（不可恢复）">🗑 清空历史</button>
             <button class="btn small" id="btn-refresh-history">刷新</button>
           </div>
         </div>
+        <div class="hist-filters">
+          <div class="icon-input" style="flex:2;min-width:160px"><span class="icon">🔍</span><input id="hist-q" placeholder="搜索：编号/名称/模型/提供商/状态/时间"></div>
+          <select id="hist-status" title="按状态筛选">
+            <option value="">状态：全部</option>
+            <option value="running">测试中</option>
+            <option value="done">完成</option>
+            <option value="error">出错</option>
+            <option value="canceled">已取消</option>
+          </select>
+          <select id="hist-provider" title="按提供商筛选">
+            <option value="">提供商：全部</option>
+            <option value="openai">OpenAI 兼容</option>
+            <option value="anthropic">Anthropic</option>
+            <option value="google">Google Gemini</option>
+          </select>
+          <select id="hist-size" title="每页条数">
+            <option value="10">10 条/页</option>
+            <option value="20" selected>20 条/页</option>
+            <option value="50">50 条/页</option>
+            <option value="100">100 条/页</option>
+          </select>
+          <button class="btn small" id="hist-reset" title="清空筛选条件并回到第一页">✕ 重置</button>
+        </div>
         <div class="table-wrap">
           <table class="history" id="history">
             <thead><tr>
               <th class="chk-col"><input type="checkbox" id="hist-check-all" title="全选"></th>
-              <th>#</th><th>时间</th><th>名称</th><th>提供商</th><th>模型</th><th>并发</th><th>采样</th>
-              <th>首字 ms</th><th>预填充 tok/s</th><th>解码 tok/s</th><th>状态</th><th>操作</th>
+              <th class="sortable" data-sort="id" title="点击排序">#</th>
+              <th class="sortable" data-sort="created_at" title="点击排序">时间</th>
+              <th class="sortable" data-sort="name" title="点击排序">名称</th>
+              <th class="sortable" data-sort="provider" title="点击排序">提供商</th>
+              <th class="sortable" data-sort="model" title="点击排序">模型</th>
+              <th>并发</th>
+              <th class="sortable" data-sort="samples" title="点击排序">采样</th>
+              <th class="sortable" data-sort="ttft" title="点击排序">首字 ms</th>
+              <th class="sortable" data-sort="prefill" title="点击排序">预填充 tok/s</th>
+              <th class="sortable" data-sort="decode" title="点击排序">解码 tok/s</th>
+              <th class="sortable" data-sort="status" title="点击排序">状态</th>
+              <th>操作</th>
             </tr></thead>
             <tbody></tbody>
           </table>
         </div>
+        <div class="hist-pager">
+          <button class="btn small" id="hist-prev">◀ 上一页</button>
+          <span class="hint" id="hist-page-info"></span>
+          <button class="btn small" id="hist-next">下一页 ▶</button>
+        </div>
       </div>
     </section>
   </main>
@@ -223,6 +274,12 @@
     </div>
   </div>
 
+  <!-- 右下角：一键置顶 / 置底 -->
+  <div class="scroll-fab">
+    <button class="fab" id="fab-top" title="回到顶部">⬆</button>
+    <button class="fab" id="fab-bottom" title="滚到底部">⬇</button>
+  </div>
+
   <script src="/js/app.js"></script>
 </body>
 </html>

static/js/app.js

@@ -69,6 +69,7 @@ function currentGen() {
     concurrency_levels: [...concurrencyLevelsActive].sort((a, b) => a - b),
     avoid_cache: $("#gen-avoid-cache").checked,
     warmup: $("#gen-warmup").checked,
+    read_timeout: parseInt($("#gen-timeout").value) || 1800,
   };
 }
 
@@ -201,9 +202,11 @@ function loadSelectedConfig() {
 async function testConnection() {
   const cfg = currentConfig();
   if (!cfg.api_key) { showConn(false, "请先填写 API Key"); return; }
+  // 慢接口长预处理可能等首字很久，连接测试也带上请求超时，避免误判接口不通
+  cfg.read_timeout = parseInt($("#gen-timeout").value) || 1800;
   const box = $("#conn-result");
   box.hidden = false; box.className = "conn-result";
-  box.textContent = "⏳ 正在测试连接...";
+  box.textContent = "⏳ 正在测试连接（等待首字，最长约 " + cfg.read_timeout + " 秒）...";
   try {
     const r = await api("/api/configs/test", "POST", cfg);
     if (r.ok) {
@@ -254,7 +257,8 @@ async function listModels() {
     const box = $("#model-result");
     box.hidden = false;
     box.className = "conn-result fail";
-    box.textContent = "❌ 获取模型失败：" + e.message;
+    box.textContent = "❌ 接口不通，请检查 Base URL / API Key 配置后重试";
+    console.warn("[listModels] 获取模型失败（已隐藏原始错误）:", e);
   }
 }
 
@@ -422,6 +426,7 @@ function stopTest() {
 /* ───────────────────────── 测试历史 ───────────────────────── */
 
 let histSelected = new Set();  // 勾选用于对比的测试 id
+const histState = { page: 1, pageSize: 20, q: "", status: "", provider: "", sort: "id", order: "desc" };
 
 function updateHistSelectedLabel() {
   const el = $("#hist-selected");
@@ -429,43 +434,74 @@ function updateHistSelectedLabel() {
 }
 
 async function loadHistory() {
-  const list = await api("/api/tests");
+  const p = new URLSearchParams({
+    page: histState.page, page_size: histState.pageSize,
+    q: histState.q, status: histState.status, provider: histState.provider,
+    sort: histState.sort, order: histState.order,
+  });
+  let d;
+  try {
+    d = await api("/api/tests?" + p.toString());
+  } catch (e) { return; }
+  if (!d || !d.ok) return;
+  renderHistory(d);
+}
+
+function renderHistory(d) {
+  const list = d.items || [];
   const tb = $("#history tbody");
   tb.innerHTML = "";
   const selAll = $("#hist-check-all");
   if (selAll) selAll.checked = false;
+  const hasFilter = histState.q || histState.status || histState.provider;
+  $("#hist-total").textContent = d.total ? `（共 ${d.total} 条）` : "";
   if (!list.length) {
-    tb.innerHTML = '<tr><td colspan="13" style="color:var(--muted);text-align:center">暂无测试记录</td></tr>';
+    tb.innerHTML = `<tr><td colspan="13" style="color:var(--muted);text-align:center">暂无${hasFilter ? "匹配的" : ""}测试记录${hasFilter ? "（试试调整筛选条件）" : ""}</td></tr>`;
     updateHistSelectedLabel();
-    return;
-  }
-  for (const t of list) {
-    const s = t.summary || {};
-    const g = t.gen || {};
-    const cls = s.concurrency_levels || g.concurrency_levels || [1];
-    const tr = document.createElement("tr");
-    tr.innerHTML = `
-      <td class="chk-col"><input type="checkbox" class="hist-chk" data-id="${t.id}" ${histSelected.has(t.id) ? "checked" : ""}></td>
-      <td>#${t.id}</td>
-      <td>${esc(t.created_at)}</td>
-      <td title="${esc(t.name || "")}">${esc(t.name || "—")}</td>
-      <td>${esc(PROVIDER_LABEL[t.provider] || t.provider)}</td>
-      <td>${esc(t.model)}</td>
-      <td class="num">${esc(cls.join("/"))}</td>
-      <td class="num">${fmt(s.samples_ok)}/${fmt(s.samples_total)}</td>
-      <td class="num">${fmt(s.avg_ttft_ms)}</td>
-      <td class="num">${fmt(s.avg_prefill_speed)}</td>
-      <td class="num">${fmt(s.avg_decode_speed)}</td>
-      <td><span class="status-pill ${esc(t.status)}">${STATUS_LABEL[t.status] || t.status}</span></td>
-      <td>
-        <button class="btn small" data-view="${t.id}">查看</button>
-        <button class="btn small" data-send="${t.id}" title="发送到评测站对应账号下">📤 发送</button>
-        <button class="btn small" data-xlsx="${t.id}">Excel</button>
-        <button class="btn small danger" data-del="${t.id}">删除</button>
-      </td>`;
-    tb.appendChild(tr);
+  } else {
+    for (const t of list) {
+      const s = t.summary || {};
+      const g = t.gen || {};
+      const cls = s.concurrency_levels || g.concurrency_levels || [1];
+      const tr = document.createElement("tr");
+      tr.innerHTML = `
+        <td class="chk-col"><input type="checkbox" class="hist-chk" data-id="${t.id}" ${histSelected.has(t.id) ? "checked" : ""}></td>
+        <td>#${t.id}</td>
+        <td>${esc(t.created_at)}</td>
+        <td title="${esc(t.name || "")}">${esc(t.name || "—")}</td>
+        <td>${esc(PROVIDER_LABEL[t.provider] || t.provider)}</td>
+        <td>${esc(t.model)}</td>
+        <td class="num">${esc(cls.join("/"))}</td>
+        <td class="num">${fmt(s.samples_ok)}/${fmt(s.samples_total)}</td>
+        <td class="num">${fmt(s.avg_ttft_ms)}</td>
+        <td class="num">${fmt(s.avg_prefill_speed)}</td>
+        <td class="num">${fmt(s.avg_decode_speed)}</td>
+        <td><span class="status-pill ${esc(t.status)}">${STATUS_LABEL[t.status] || t.status}</span></td>
+        <td>
+          <button class="btn small" data-view="${t.id}">查看</button>
+          <button class="btn small" data-send="${t.id}" title="发送到评测站对应账号下">📤 发送</button>
+          <button class="btn small" data-xlsx="${t.id}">Excel</button>
+          <button class="btn small danger" data-del="${t.id}">删除</button>
+        </td>`;
+      tb.appendChild(tr);
+    }
   }
   updateHistSelectedLabel();
+
+  // 分页信息
+  const totalPages = Math.max(1, d.total_pages || 1);
+  $("#hist-page-info").textContent = `第 ${d.page} / ${totalPages} 页 · 共 ${d.total} 条`;
+  $("#hist-prev").disabled = d.page <= 1;
+  $("#hist-next").disabled = d.page >= totalPages;
+  histState.page = d.page;
+  updateSortIndicators();
+}
+
+function updateSortIndicators() {
+  $$("#history thead th.sortable").forEach((th) => {
+    th.classList.remove("asc", "desc");
+    if (th.dataset.sort === histState.sort) th.classList.add(histState.order);
+  });
 }
 
 /* ───────────────────────── 多测试对比 ───────────────────────── */
@@ -1064,8 +1100,72 @@ function bind() {
 
   $("#btn-clear-console").addEventListener("click", clearConsole);
   $("#btn-export-log").addEventListener("click", exportCurrentLog);
-  $("#btn-refresh-history").addEventListener("click", loadHistory);
+  $("#btn-refresh-history").addEventListener("click", () => { histState.page = 1; loadHistory(); });
   $("#btn-compare").addEventListener("click", openCompare);
+  $("#btn-export-history").addEventListener("click", exportAllHistory);
+  $("#btn-clear-history").addEventListener("click", clearAllHistory);
+
+  // 左侧三块：点击标题折叠/展开（状态存 localStorage）
+  const COLLAPSE_KEY = "llmst.collapsed";
+  let collapsedSections = new Set();
+  try { collapsedSections = new Set(JSON.parse(localStorage.getItem(COLLAPSE_KEY) || "[]")); } catch (e) {}
+  document.querySelectorAll(".card.collapsible").forEach((sec) => {
+    if (collapsedSections.has(sec.id)) sec.classList.add("collapsed");
+    const head = sec.querySelector(".collapse-head");
+    if (head) head.addEventListener("click", () => {
+      const on = sec.classList.toggle("collapsed");
+      if (on) collapsedSections.add(sec.id); else collapsedSections.delete(sec.id);
+      try { localStorage.setItem(COLLAPSE_KEY, JSON.stringify([...collapsedSections])); } catch (e) {}
+    });
+  });
+
+  // 历史：筛选 / 搜索 / 每页条数 / 分页 / 重置
+  let histQTimer = null;
+  $("#hist-q").addEventListener("input", () => {
+    clearTimeout(histQTimer);
+    histQTimer = setTimeout(() => {
+      histState.q = $("#hist-q").value.trim();
+      histState.page = 1;
+      loadHistory();
+    }, 300);
+  });
+  $("#hist-status").addEventListener("change", () => {
+    histState.status = $("#hist-status").value;
+    histState.page = 1;
+    loadHistory();
+  });
+  $("#hist-provider").addEventListener("change", () => {
+    histState.provider = $("#hist-provider").value;
+    histState.page = 1;
+    loadHistory();
+  });
+  $("#hist-size").addEventListener("change", () => {
+    histState.pageSize = parseInt($("#hist-size").value) || 20;
+    histState.page = 1;
+    loadHistory();
+  });
+  $("#hist-prev").addEventListener("click", () => {
+    if (histState.page > 1) { histState.page--; loadHistory(); }
+  });
+  $("#hist-next").addEventListener("click", () => {
+    if (!$("#hist-next").disabled) { histState.page++; loadHistory(); }
+  });
+  $("#hist-reset").addEventListener("click", resetHistoryFilters);
+
+  // 历史：点击表头升降排序
+  $("#history thead").addEventListener("click", (e) => {
+    const th = e.target.closest("th.sortable");
+    if (!th) return;
+    const key = th.dataset.sort;
+    if (histState.sort === key) {
+      histState.order = histState.order === "asc" ? "desc" : "asc";
+    } else {
+      histState.sort = key;
+      histState.order = (key === "id" || key === "created_at") ? "desc" : "asc";
+    }
+    histState.page = 1;
+    loadHistory();
+  });
 
   $("#history tbody").addEventListener("click", (e) => {
     const v = e.target.closest("[data-view]");
@@ -1108,6 +1208,57 @@ function bind() {
   $("#dt-send-eval").addEventListener("click", () => { if (window.__detail) sendToEval(window.__detail.id); });
 
   document.addEventListener("keydown", (e) => { if (e.key === "Escape") { hideModelPick(); closeDetail(); closeCompare(); } });
+
+  // 右下角一键置顶 / 置底
+  $("#fab-top").addEventListener("click", () => window.scrollTo({ top: 0, behavior: "smooth" }));
+  $("#fab-bottom").addEventListener("click", () =>
+    window.scrollTo({ top: document.documentElement.scrollHeight, behavior: "smooth" }));
+}
+
+/* ───────────────────────── 历史筛选工具 ───────────────────────── */
+
+function resetHistoryFilters() {
+  histState.q = ""; histState.status = ""; histState.provider = "";
+  histState.sort = "id"; histState.order = "desc";
+  histState.page = 1;
+  $("#hist-q").value = "";
+  $("#hist-status").value = "";
+  $("#hist-provider").value = "";
+  loadHistory();
+}
+
+async function clearAllHistory() {
+  if (!confirm("确定清空全部测试历史？\n所有测试记录、采样数据与日志将被永久删除，无法恢复！")) return;
+  if (!confirm("再次确认：真的要清空全部历史记录吗？")) return;
+  const r = await api("/api/tests", "DELETE");
+  if (r.ok) {
+    histSelected.clear();
+    histState.page = 1;
+    toast("✅ 历史记录已全部清空");
+    loadHistory();
+    updateHistSelectedLabel();
+  } else {
+    toast("❌ 清空失败：" + (r.error || "未知错误"));
+  }
+}
+
+async function exportAllHistory() {
+  try {
+    const resp = await fetch("/api/tests/export");
+    if (!resp.ok) throw new Error("HTTP " + resp.status);
+    const blob = await resp.blob();
+    const cd = resp.headers.get("Content-Disposition") || "";
+    const m = cd.match(/filename\"?=?\"?([^";]+)\"?/);
+    const name = m ? m[1] : "llm_speed_tests.json";
+    const url = URL.createObjectURL(blob);
+    const a = document.createElement("a");
+    a.href = url; a.download = name;
+    document.body.appendChild(a); a.click(); a.remove();
+    URL.revokeObjectURL(url);
+    toast("✅ 已导出全部历史（共 " + (blob.size / 1024).toFixed(1) + " KB）");
+  } catch (e) {
+    toast("❌ 导出失败：" + e.message);
+  }
 }
 
 /* ───────────────────────── 初始化 ───────────────────────── */

tester.py

@@ -77,6 +77,9 @@ class TestRunner(threading.Thread):
                  % (" / ".join(str(x) for x in lengths), max_tokens,
                     " / ".join(str(x) for x in concurrency_levels), n,
                     "开" if warmup else "关", "开" if avoid_cache else "关"))
+        to = self.gen.get("read_timeout") or "默认(1800)"
+        self.log("INFO", "请求超时: 连接 %s s | 等待首字/预处理 %s s（接口慢可在左侧调大）"
+                 % (self.gen.get("connect_timeout") or 60, to))
 
         ratio = self._calibrate()
         self.ratio = ratio
@@ -157,7 +160,9 @@ class TestRunner(threading.Thread):
         """
         from concurrent.futures import ThreadPoolExecutor
 
-        gen_opt = {"max_tokens": max_tokens, "avoid_cache": avoid_cache}
+        gen_opt = {"max_tokens": max_tokens, "avoid_cache": avoid_cache,
+                   "connect_timeout": self.gen.get("connect_timeout"),
+                   "read_timeout": self.gen.get("read_timeout")}
 
         def worker(idx):
             prompt = self._finalize_prompt(base_prompt)  # 每流独立随机前缀，避免共享缓存
@@ -200,6 +205,10 @@ class TestRunner(threading.Thread):
         total_ms = max((batch_end - batch_start) * 1000.0, 0.1)
         prefill = (total_prompt / (ttft_ms / 1000.0)) if total_prompt else None
         decode = (total_output / (decode_ms / 1000.0)) if total_output else None
+        # 防“幻影 token”：所有流都没收到任何流式文本但 usage 报了输出 token（如推理用满 max_tokens），
+        # 整批解码时间无意义，不报数百万的荒谬解码速度
+        if total_output and decode is not None and not total_ochars:
+            decode = None
 
         agg = {
             "concurrency": concurrency,
@@ -236,7 +245,9 @@ class TestRunner(threading.Thread):
         self.log("INFO", "正在校准 token/字符 比例（发送小探测请求）...")
         try:
             m = lp.call_stream(self.cfg, probe,
-                               {"max_tokens": 8, "avoid_cache": False},
+                               {"max_tokens": 8, "avoid_cache": False,
+                                "connect_timeout": self.gen.get("connect_timeout"),
+                                "read_timeout": self.gen.get("read_timeout")},
                                log=lambda lv, msg: self.log(lv, msg),
                                should_stop=self.should_stop)
             pt = m.get("prompt_tokens") or 0