公告TokenPrice 上线：Claude Code、Codex 怎么用最便宜，三条路实时比价加入交流群
TokenPriceAI 编程省钱比价
商务合作

SHUAI API 怎么样

api.shuaiapi.comNew API在线更新于 1 小时前
去官网

SHUAI API（api.shuaiapi.com）是一家 API 中转站，用 New API 系统搭建，充值 ¥1 = $1，Claude 模型最低 1.15x（官方 1 美元用量花 ¥1.15，约官方价的 1.7 折），在 1,932 家里排第 1,344，近期公开接口可用率 100%。以上是公开标价，我们没有实测模型真假和速度。

Claude 最低倍率1.15x官方 1 美元用量花 ¥1.15
充值比例¥1 = $1官方约 ¥6.68 = $1
公开接口可用率100%只说明站点在线，不代表模型质量
模型数77共 177 条报价
RATIO

各模型系列的倍率和名次

倍率 = 官方每 1 美元的用量在这家花多少人民币；同一系列不同分组差很多

系列最低最高分组折合官方榜单名次
Claude 中转站榜1.15x2x1.7 折1,344 / 1,932
GPT 中转站榜0.2x0.35x0.3 折771 / 2,117
Gemini 中转站榜0.4x0.4x0.6 折457 / 1,418
DeepSeek 中转站榜0.35x1x0.52 折579 / 1,552
GLM 中转站榜0.35x1x0.52 折468 / 1,331
Kimi 中转站榜0.35x1x0.52 折434 / 1,224
Qwen 中转站榜0.35x1x0.52 折224 / 804
MODELS

SHUAI API 常用模型价格

人民币 / 百万 token，每个模型取这家最便宜的分组（免费、体验分组不算）

模型分组输入输出
gpt-5.6-solAvailability is not guaranteed for this group; cache hit rates depend entirely on luck.¥0.32¥1.6
gpt-6-astraAvailability is not guaranteed for this group; cache hit rates depend entirely on luck.¥0.8¥4
gpt-5.6-terraAvailability is not guaranteed for this group; cache hit rates depend entirely on luck.¥0.16¥0.96
gpt-6.1-solAvailability is not guaranteed for this group; cache hit rates depend entirely on luck.¥0.16¥0.8
gpt-6-solAvailability is not guaranteed for this group; cache hit rates depend entirely on luck.¥0.16¥0.8
claude-opus-4-8Not self-operated; data may be inflated or have other issues. Please test before integration; no guarantee is provided.¥4.5¥22.5
claude-opus-5Not self-operated; data may be inflated or have other issues. Please test before integration; no guarantee is provided.¥4.5¥22.5
claude-sonnet-5Not self-operated; data may be inflated or have other issues. Please test before integration; no guarantee is provided.¥1.8¥9
claude-opus-4-7Not self-operated; data may be inflated or have other issues. Please test before integration; no guarantee is provided.¥4.5¥22.5
claude-opus-4-6Not self-operated; data may be inflated or have other issues. Please test before integration; no guarantee is provided.¥4.5¥22.5
claude-sonnet-4-6Not self-operated; data may be inflated or have other issues. Please test before integration; no guarantee is provided.¥2.7¥13.5
claude-fable-5Not self-operated; data may be inflated or have other issues. Please test before integration; no guarantee is provided.¥9¥45
claude-opus-5-5Not self-operated; data may be inflated or have other issues. Please test before integration; no guarantee is provided.¥3.6¥18
claude-fable-5-1Not self-operated; data may be inflated or have other issues. Please test before integration; no guarantee is provided.¥9¥45
claude-sonnet-5-5Non-proprietary, high stability, fully verified account pool. You can report upstream model degradation directly to me. Please test before integration.¥2¥10
gpt-5.6-lunaOfficial API keys only, high cache hit rates, and content moderation enabled. No compensation if the service goes down.¥0.06¥0.36
gpt-5.5Official API keys only, high cache hit rates, and content moderation enabled. No compensation if the service goes down.¥1.5¥9
claude-haiku-4-5-20251001Not self-operated; data may be inflated or have other issues. Please test before integration; no guarantee is provided.¥0.9¥4.5
deepseek-v4.1-flashAvailability has been self-tested. The model is real. For caching, please review the performance page before using it.¥0.7¥2.8
gpt-6-lunaOfficial API keys only, high cache hit rates, and content moderation enabled. No compensation if the service goes down.¥0.03¥0.15
codex-auto-reviewAvailability is not guaranteed for this group; cache hit rates depend entirely on luck.¥0.06¥0.36
kimi-k3Availability has been self-tested. The model is real. For caching, please review the performance page before using it.¥7¥35
glm-5.3Availability has been self-tested. The model is real. For caching, please review the performance page before using it.¥2.8¥9.8
deepseek-v4-proAvailability has been self-tested. The model is real. For caching, please review the performance page before using it.¥3.15¥9.45
gpt-6.1-terraWhether degradation occurs depends on detection; we will do our best to ensure quality.¥0.2¥1
gemini-3.8-flashSupports Gemini4, high concurrency, and Antigravity, with slightly lower cache hit rates. Recent samples show relatively high cache hit rates with some upstream providers when OpenCode uses Chat Completions (/v1/chat/completions) and Pi uses Responses (/v1/responses). Consistently reusing context helps improve cache hit rates, but Responses cache hit rates remain relatively low with some upstream providers. Refer to your actual usage logs for the most accurate picture.¥0.3¥1.5
glm-5.3-flashAvailability has been self-tested. The model is real. For caching, please review the performance page before using it.¥0.28¥0.98
claude-haiku-5-5Non-proprietary, high stability, fully verified account pool. You can report upstream model degradation directly to me. Please test before integration.¥0.1¥0.5
glm-5.2Availability has been self-tested. The model is real. For caching, please review the performance page before using it.¥2.8¥9.8
gemini-4-argonSupports Gemini4, high concurrency, and Antigravity, with slightly lower cache hit rates. Recent samples show relatively high cache hit rates with some upstream providers when OpenCode uses Chat Completions (/v1/chat/completions) and Pi uses Responses (/v1/responses). Consistently reusing context helps improve cache hit rates, but Responses cache hit rates remain relatively low with some upstream providers. Refer to your actual usage logs for the most accurate picture.¥0.8¥4
SETUP

怎么接到 Claude Code、Codex

按公开网址写的通用写法；这家具体支持哪种接口，以它的文档为准

~/.claude/settings.json（Claude Code）
{
  "env": {
    "ANTHROPIC_BASE_URL": "https://api.shuaiapi.com",
    "ANTHROPIC_AUTH_TOKEN": "这家后台生成的 Key"
  }
}
~/.codex/config.toml（Codex）
model_provider = "relay"

[model_providers.relay]
name = "SHUAI API"
base_url = "https://api.shuaiapi.com/v1"
env_key = "RELAY_API_KEY"
wire_api = "responses"
先小额测试

我们没有实测这家的模型真假和速度。充值前先确认网址，再按 中转站怎么测真假、防跑路 测一遍。

SIMILAR

Claude 价格相近的中转站

榜单上名次挨着的几家

完整榜单
FAQ

常见问题

SHUAI API 怎么样、靠谱吗？
本站只记录公开数据，没有充值实测：标价上Claude 1.15x（1.7 折）、GPT 0.2x（0.3 折）、Gemini 0.4x（0.6 折）、DeepSeek 0.35x（0.52 折），公开接口可用率 100%。模型真假、速度和售后要自己测，先充小额，方法见 中转站怎么测真假、防跑路。
SHUAI API 官网是哪个？
本站记录的网址是 api.shuaiapi.com（https://api.shuaiapi.com）。中转站换域名很常见，注册和充值前请再确认一次。
SHUAI API 的倍率是多少？
Claude 1.15x – 2x；GPT 0.2x – 0.35x；Gemini 0.4x；DeepSeek 0.35x – 1x；GLM 0.35x – 1x；Kimi 0.35x – 1x；Qwen 0.35x – 1x。倍率 = 官方每 1 美元用量在这家花多少人民币，官方直接买约 ¥6.68。同一家不同分组价格和渠道不同，最低的分组不一定稳定。
SHUAI API 能接 Claude Code 吗？
标价上有 Claude 模型（1.15x 起）。Claude Code 走 Claude 格式的 /v1/messages 接口，设置 ANTHROPIC_BASE_URL 和 Key 就能接，但这家是否支持、支持哪些分组，要看它的文档或问客服。配置步骤见 Claude Code 接中转 API 教程。
SHUAI API 能接 Codex 吗？
标价上有 GPT 模型（0.2x 起）。Codex 只支持 Responses 接口，这家要提供 /v1/responses 才能用，买前先确认。配置见 Codex 怎么用最省钱。
和 SHUAI API 价格差不多的中转站有哪些？
Claude 榜单上名次相近的有 0AIT（1.09x）、IT Project（1.09x）、Joint（1.1x）、小鹿AI（1.1x）、万法AI（1.1x）。完整排名见 Claude 中转站价格榜。

一起把 AI 编程的钱花在刀刃上

交流群里聊 Claude Code、Codex 怎么用更省、哪家中转站最近稳；看到价格错了、店铺跑路了，也欢迎来说一声。

加入交流群

聊 Claude Code、Codex 怎么用更省，反馈价格错误