← 返回 Skills 市场

Edge AI PC Benchmark

Name: Edge AI PC Benchmark
Author: ai-gaoqian

作者 ai-gaoqian · GitHub ↗ · v1.0.0 · MIT-0

cross-platform ✓ 安全检测通过

总下载

当前安装

版本数

在 OpenClaw 中安装

/install edge-ai-pc-benchmark

功能描述

Compare local AI PC performance across 10+ chips, see tokens/sec for LLMs, model compatibility, developer tools, and WWDC 2026 Apple Silicon previews.

使用说明 (SKILL.md)

\r \r

端侧 AI PC 跑分榜 (Edge AI PC Benchmark)\r

概述\r

\r 最全面的端侧 AI PC 性能对比引擎。覆盖 10+ 芯片（NVIDIA RTX Spark、Apple M4 Ultra、AMD Ryzen AI 400、Intel Nova Lake、Qualcomm X Elite Gen 2 等）的本地大模型推理性能，提供 tokens/sec 实测数据、模型兼容性矩阵、开发者工具链选购指南。\r \r

适用场景\r

AI PC 购机指南（哪台能跑 Llama 70B？预算多少？）\r
本地 LLM 性能对比（tokens/sec 实测排行）\r
芯片技术参数对比（TOPS/内存带宽/功耗）\r
模型兼容性查询（哪些模型能在哪个芯片上用哪种量化格式运行？）\r
开发者工具链推荐（Ollama / LM Studio / llama.cpp / vLLM）\r
WWDC 2026 Apple Silicon AI 前瞻\r \r

输入参数\r

\r | 参数 | 类型 | 必填 | 说明 |\r |------|------|------|------|\r | chip | string | 否 | RTX Spark / M4 Ultra / Ryzen AI 400 / Nova Lake / X Elite |\r | model_size | string | 否 | 7B / 8B / 13B / 34B / 70B / 200B+ |\r | use_case | string | 否 | buy / benchmark / compatibility / dev-tools / wwdc |\r \r

定价\r

\r | 项目 | 值 |\r |------|-----|\r | 单价 | ¥0.50 / 次 |\r | 支付协议 | alipay_ 短链协议 |\r | 网关地址 | http://8.145.54.67:3000 |\r | 技能路径 | /skill/edge-ai-pc-benchmark |\r

安全使用建议

Before installing, note that some workflows are intentionally powerful: moderation commands can affect real users and skills, and the autoreview helper may run nested review with full filesystem authority unless configured otherwise. Use these only in the intended ClawHub repository context and review confirmation prompts carefully.

能力评估

ℹ Purpose & Capability

The artifact content is oriented around repository maintenance, ClawHub moderation workflows, UI proof generation, autoreview, and Convex development guidance; these capabilities fit the stated purposes.

ℹ Instruction Scope

Some workflows involve high-impact actions such as moderator commands and nested code review execution, but the instructions disclose them and generally require explicit targets, reasons, confirmation, auth, and verification.

✓ Install Mechanism

No suspicious install-time behavior, package self-modification, hidden bootstrap, or dependency-install side channel was evident in the inspected skill files.

ℹ Credentials

The autoreview helper can run nested Codex review with full-access sandbox bypass by default, which is powerful but disclosed and tied to the review workflow with an opt-out option.

✓ Persistence & Privilege

No hidden persistence, background daemon setup, credential harvesting, shell profile modification, or privilege escalation mechanism was found.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install edge-ai-pc-benchmark
安装完成后，直接呼叫该 Skill 的名称或使用 /edge-ai-pc-benchmark 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v1.0.0

Edge AI PC Benchmark 1.0.0 - Initial release featuring comprehensive benchmarking across 10+ Edge AI PC chips, including NVIDIA RTX Spark, Apple M4 Ultra, Ryzen AI 400, and Intel Nova Lake. - Provides real-world local LLM performance metrics (tokens/sec) and detailed compatibility matrix. - Includes developer toolchain recommendations and a preview of WWDC 2026 Apple Silicon AI. - Supports queries by chip, model size, and use case. - Includes an AI PC buying guide and pricing info (¥0.50 per use).

元数据

Slug edge-ai-pc-benchmark

版本 1.0.0

许可证 MIT-0

累计安装 1

当前安装数 1

历史版本数 1

常见问题