Ai Model Comparison

AI 模型对比指南 - 深度对比主流模型,帮你选最合适的。适合:选型困难、成本优化。

MIT-0 · Free to use, modify, and redistribute. No attribution required.
0 · 30 · 1 current installs · 1 all-time installs
MIT-0
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
Purpose & Capability
Name and description match the SKILL.md content (model comparisons, pricing, scenarios). No unrelated binaries, env vars, or config paths are requested.
Instruction Scope
SKILL.md contains only static guidance, tables, YAML examples, pricing and contact info — it does not instruct the agent to read files, access environment variables, call external endpoints, or execute commands.
Install Mechanism
No install spec or code files are present; this is instruction-only and does not write to disk or download code.
Credentials
No credentials, tokens, or environment variables are requested; the guidance does not require secret access.
Persistence & Privilege
Skill is not always-enabled and does not request persistent privileges or modify agent/system settings.
Assessment
This skill is a static, instruction-only comparison guide and appears coherent with its stated purpose. It does not request credentials or install code, so the direct security risk is low. Consider: (1) pricing and model capabilities may be out-of-date or inaccurate — verify with official provider docs before making purchases or architecture decisions; (2) the SKILL.md includes contact/payment info (WeChat/Telegram) — be cautious about sharing payment or sensitive information with unknown individuals; (3) since it's instruction-only, it cannot perform actions for you — any automated integration would require separate tooling and credentials. If you need automated comparisons or API access, prefer reputable provider SDKs and only grant the minimum credentials required.

Like a lobster shell, security has layers — review code before you run it.

Current versionv1.0.3
Download zip
latestvk9776ytd7zp59h10qfvs98b4xh82yf4x

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

Runtime requirements

🔬 Clawdis

SKILL.md

AI 模型对比指南

主流模型深度对比,选最合适的。

快速选择

需求推荐模型理由
日常对话DeepSeek V3便宜、中文好
编程辅助Claude Sonnet代码能力强
长文档分析GPT-4-Turbo128K 上下文
免费使用GLM-4每月 100 万 tokens
推理任务DeepSeek R1推理能力强
多模态GPT-4o图片+文字

价格对比

输入价格(¥/百万 tokens)

模型价格
DeepSeek V3¥0.27
GLM-4¥0.1(免费额度内)
Claude Haiku¥0.7
GPT-4o-mini¥1.75
DeepSeek R1¥1.35
Claude Sonnet¥21
GPT-4o¥18
Claude Opus¥35

输出价格(¥/百万 tokens)

模型价格
DeepSeek V3¥1.08
GLM-4¥0.1(免费额度内)
Claude Haiku¥3.5
GPT-4o-mini¥10.5
DeepSeek R1¥5.4
Claude Sonnet¥105
GPT-4o¥108
Claude Opus¥175

结论:DeepSeek 最便宜,GLM 免费额度最大

性能对比

编程能力

模型评分特点
Claude Sonnet⭐⭐⭐⭐⭐代码质量高,解释清晰
GPT-4o⭐⭐⭐⭐⭐综合能力强
DeepSeek V3⭐⭐⭐⭐性价比高
GLM-4⭐⭐⭐⭐中文代码好

中文能力

模型评分特点
DeepSeek V3⭐⭐⭐⭐⭐国产,中文优秀
GLM-4⭐⭐⭐⭐⭐智谱,中文优秀
Claude⭐⭐⭐⭐翻译感
GPT-4o⭐⭐⭐⭐英文原生

推理能力

模型评分特点
DeepSeek R1⭐⭐⭐⭐⭐推理专精
Claude Opus⭐⭐⭐⭐⭐综合推理强
GPT-4o⭐⭐⭐⭐⭐综合能力强
DeepSeek V3⭐⭐⭐⭐性价比高

长文本

模型上下文特点
GPT-4-Turbo128K支持
Claude200K支持
GLM-4-Long128K支持
DeepSeek64K支持

免费额度对比

模型免费额度条件
GLM-4100 万 tokens/月新用户永久
DeepSeek V310 次/天注册即有
Claude$5 注册送新用户
GPT-4o$5 注册送新用户

结论:GLM 免费额度最大

场景推荐

个人日常

主模型:DeepSeek V3(便宜)
备用:GLM-4(免费额度)

月成本:¥0-10

编程开发

主模型:Claude Sonnet(代码质量高)
简单任务:DeepSeek V3(便宜)

月成本:¥50-200

内容创作

主模型:DeepSeek V3(便宜、中文好)
审核:Claude(高质量)}

月成本:¥10-50

企业应用

主模型:GPT-4o(稳定)
降级:DeepSeek V3(省钱)

月成本:¥500-5000

省钱策略

分层使用

models:
  simple: "deepseek-chat"      # ¥0.27/百万
  normal: "glm-4"              # 免费(额度内)
  complex: "gpt-4o-mini"       # ¥1.75/百万
  critical: "claude-sonnet"    # ¥21/百万(关键任务)

自动降级

fallback:
  - claude-sonnet
  - gpt-4o-mini
  - deepseek-chat
  - glm-4

缓存策略

cache:
  enabled: true
  ttl: 86400  # 24 小时
  # 节省 30-50% 成本

需要帮助?

  • 模型选型:¥99
  • 成本优化:¥299
  • 企业方案:¥999

联系:微信 yang1002378395 或 Telegram @yangster151


创建:2026-03-14

Files

1 total
Select a file
Select a file to preview.

Comments

Loading comments…