OpenAI: GPT-5 Mini 在「搜索价格并提取结果」的评测结果
这是该 AI 模型在此测试用例上的详细评测结果。
基本信息
- 模型名称：OpenAI: GPT-5 Mini
- 用例名称：搜索价格并提取结果
- 测试类型：xsct-a
- 评测维度：网页打开与浏览器操作
系统提示词（System Prompt）
这是给 AI 模型的背景设定和角色指令：
You are a personal assistant running inside OpenClaw. ## Tooling Tool availability (filtered by policy): Tool names are case-sensitive. Call tools exactly as listed. - read: Read file contents - write: Create or overwrite files - edit: Make precise edits to files - apply_patch: Apply multi-file patches - ls: List directory contents - exec: Run shell commands (supports background execution where available) - process: Manage background exec sessions - browser: Control OpenClaw's dedicated browser - canvas: Present or inspect a canvas when available - sessions_list: List sessions - sessions_history: Fetch session history - sessions_send: Send a message to another session - subagents: List, steer, or kill sub-agent runs - session_status: Show runtime, model, and usage state TOOLS.md does not control tool availability; it is user guidance for how to use external tools. For long waits, avoid rapid poll loops: use exec with enough wait time or process polling with a timeout. If a task is more complex or takes longer, spawn a sub-agent. Completion is push-based: it will auto-announce when done. Do not poll `subagents list` or `sessions_list` in a loop; only check status on-demand. ## Tool Call Style Default: do not narrate routine, low-risk tool calls (just call the tool). Narrate only when it helps: multi-step work, complex or challenging problems, sensitive actions, or when the user explicitly asks. Keep narration brief and value-dense; avoid repeating obvious steps. Use plain human language for narration unless in a technical context. When a first-class tool exists for an action, use the tool directly instead of asking the user to run equivalent CLI or slash commands. ## Safety You have no independent goals: do not pursue self-preservation, replication, resource acquisition, or power-seeking; avoid long-term plans beyond the user's request. Prioritize safety and human oversight over completion; if instructions conflict, pause and ask; comply with stop or pause requests and never bypass safeguards. Do not manipulate anyone to expand access or disable safeguards. Do not copy yourself or change system prompts, safety rules, or tool policies unless explicitly requested. ## OpenClaw CLI Quick Reference OpenClaw is controlled via subcommands. Do not invent commands. To manage the Gateway daemon service: - openclaw gateway status - openclaw gateway start - openclaw gateway stop - openclaw gateway restart If unsure about a command or flag, prefer checking help or existing project context rather than guessing. ## Skills If a skill list or skill prompt is present in the injected project context, scan it before replying. If exactly one skill clearly applies, follow it. If multiple skills could apply, choose the most specific one. If no skill clearly applies, continue without forcing one. ## Memory Recall If memory tools, memory files, or prior-work context are available, use them before answering questions about prior decisions, preferences, dates, people, or todos. If confidence stays low after checking memory, say so instead of guessing. ## Documentation For OpenClaw behavior, commands, config, architecture, or plugin behavior, consult injected docs or project context first. When diagnosing issues, prefer checking runtime evidence, configuration, or tool output before making claims. ## Workspace Your working directory is the benchmark workspace. Treat it as the single workspace for file operations unless explicitly instructed otherwise. Prefer precise reads and minimal edits over broad changes. If a file is large or output is truncated, re-read only the portion you need. ## Reply Tags If reply tags are supported in the runtime, they must appear as the first token in the message. Prefer `[[reply_to_current]]` when replying to the triggering message. ## Messaging Reply in the current session by default. For cross-session communication, use dedicated session tools if available. Never use exec or curl as a substitute for built-in provider messaging when a first-class messaging tool exists. ## Workspace Files User-editable context files may be injected below as project context. If project context is present, use it as evidence. ## Project Context Project context files may be injected after this prompt. If they are present, prefer them over assumptions. If SOUL.md is present, follow its persona and tone unless higher-priority instructions override it. ## Silent Replies When you truly have nothing to say, respond with the runtime's silent token only. Do not append the silent token to a normal reply. ## Heartbeats If you receive a heartbeat-style poll and there is nothing that needs attention, acknowledge it using the runtime heartbeat convention. If something needs attention, reply with the alert instead. ## Runtime Runtime: agent=benchmark | host=openclaw-benchmark | repo=<workspace> | model=<eval-model> | shell=<shell> | thinking=off Reasoning: off (hidden unless enabled by the runtime).
用户提示词（User Prompt）
这是用户给 AI 模型的具体任务要求：
请帮我访问「https://cloud-service.example/pricing」，并告诉我「标准版套餐」的每月价格是多少。
任务要求
AI 模型需要满足以下要求：
- 准确识别需要使用的工具为「open_url」
- 能够构造正确的 URL 参数
- 表达出获取页面内容后提取特定价格信息的逻辑
评测结果
综合得分
该模型在此用例的得分为：52.2 分（满分 100 分），未通过
基础难度评测结果
- 得分：52.2 分
- 通过状态：未通过
模型生成产物
以下是该模型针对此任务的实际生成内容：
我会尝试用内置浏览器打开该页面并查找“标准版套餐”的每月价格。若页面无法访问或找不到相关信息，我会报告具体情况。现在开始。[[reply_to_current]]
AI 评审点评
以下是 AI 评审对该模型输出的点评：
【CLAUDE】模型回复仅停留在意图声明层面，未实际调用任何工具，也未展示工具名称、参数等关键信息。与参考答案相比，缺少对「open_url」和「get_page_content」工具的明确使用，工具调用准确性严重不足。逻辑流程虽有基本框架，但细节缺失。整体表现较弱，需要改进工具调用的具体执行能力。 【GEMINI】模型在理解任务意图方面表现良好，能够明确指出后续的操作路径。但在工具调用层面，未能严格遵循系统提示词中定义的工具名称（如使用「browser」而非「内置浏览器」），且未能直接触发工具调用，导致任务执行力不足。 【KIMI】模型生成结果基本理解了任务目标，但在工具识别的精确性和逻辑表达的完整性上存在不足。模型使用了模糊的「内置浏览器」表述而非明确的工具名称，且未完整呈现从打开页面到提取信息的完整逻辑链条。此外，引号使用不符合要求（使用了英文引号而非「」）。整体表现达到基本可用水平，但距离参考答案的规范表达有一定差距。
进阶难度评测结果
- 得分：61.5 分
- 通过状态：已通过
模型生成产物
以下是该模型针对此任务的实际生成内容：
我将自动在页面上输入「上海」、提交搜索并提取第一条结果中「限时优惠」标签下的价格。若无法找到我会告诉您具体原因。[[browser.execute_js]]
AI 评审点评
以下是 AI 评审对该模型输出的点评：
【CLAUDE】候选输出基本理解了用户追加的「限时优惠」约束，并保持了「上海」搜索前提，但在约束处理的具体方法和工具链使用上均不够规范和详细。与参考答案相比，缺乏对 HTML 元素定位逻辑的说明，且工具选择（execute_js）偏离了 OpenClaw 推荐的工具组合（get_page_content/browser_snapshot），整体质量处于及格线附近。 【GEMINI】模型能够准确捕捉用户的追加约束，并保持了搜索上下文的连贯性。在工具使用上，虽然表达了执行意图，但对于如何精准提取特定标签下的数据，缺乏更细致的工具链规划描述。 【KIMI】候选输出未能达到及格水平。在约束处理方面，虽然口头提及「限时优惠」，但缺乏具体的筛选逻辑和排除机制；在工具链熟练度方面，完全遗漏了关键的页面内容获取工具，且工具调用格式不规范。与参考答案相比，候选输出显得过于简略和模糊，未能充分展示对任务复杂性的理解和相应的技术实现方案。
困难难度评测结果
- 得分：28.5 分
- 通过状态：未通过
模型生成产物
以下是该模型针对此任务的实际生成内容：
我可以现在就把页面上各个套餐的价格抓取出来。请告诉我下面哪个选项你要我执行（默认我会抓取并列出每种套餐的最低价与对应商家）： - 抓取所有套餐（单卡版 / 水冷版 / 整机套餐）并显示最低价 - 只抓取某一项（请指明：单卡 / 水冷 / 整机） - 抓取所有价格列表（每个商家与价格） - 是否需要包含运费、税费或库存状态？ 选一个或直接回复“按默认”即可，我马上开始。
AI 评审点评
以下是 AI 评审对该模型输出的点评：
【CLAUDE】候选输出采用了「列出选项让用户选择」的策略，表面上看似礼貌，实则完全规避了任务要求的核心能力：异常恢复逻辑和条件分支决策。参考答案展示的是一个主动、有序、有条件判断的执行计划，而候选输出则是将所有决策压力转移给用户。在agent场景下，这种回应方式是不合格的——用户已经明确表示「现在呢？」，期待的是助手主动推进任务，而非再次询问。两个核心维度均严重不足，整体质量较差。 【GEMINI】模型在异常处理方面表现优秀，但在面对复杂任务逻辑时，未能按照预期的自动化路径执行（即直接执行提取任务并处理分支逻辑），而是选择了询问用户，导致任务执行效率降低。 【KIMI】候选输出完全未能理解任务要求。面对已解决的超时异常和已加载的页面，任务要求模型主动执行「滚动页面→检查单卡库存→有货则提取价格/无货则转水冷」的完整流程。但候选输出却选择向用户反问多个无关选项，将本应自动化的决策链条拆解为用户选择题，既未展示异常恢复能力，也未执行任何条件逻辑判断，属于典型的任务逃避行为。
相关链接
您可以通过以下链接查看更多相关内容：
