点 token，直达省钱榜
省下来的，都是你的。
覆盖国内外主流平台，全面比价各家热门模型在不同渠道的价格、免费额度与限时活动，挑出更划算的，第一时间提醒您。免费订阅。
● 实时价格·选完即对比
| 服务商 | 支持模型 | 官网倍率/价格 | 状态 | |
|---|---|---|---|---|
| DeepSeek 官方API · V4 Flash / Pro | DeepSeek V4 Flash / Pro | 1.0×每1M输出token | 官方低价 | 去看看 → |
| 硅基流动SiliconFlow · 已备案 | Qwen / GLM / DeepSeek | 免费模型每1M输出token | 0× 模型 | 去看看 → |
| 阿里百炼阿里云 · 企业级 | 通义千问 / 万相 | 免费额度每1M输出token | 新户试用 | 去看看 → |
| 智谱 GLM智谱AI · 已备案 | GLM 系列 | 免费额度每1M输出token | 新户资源包 | 去看看 → |
| MiniMax稀宇科技 · 已备案 | MiniMax M3 / M2 | 1.0×每1M输出token | 套餐价 | 去看看 → |
| 火山方舟字节跳动 · 企业级 | 豆包 / DeepSeek / Kimi | 1.0×每1M输出token | 企业级 | 去看看 → |
| OpenRouter聚合 594 模型 | GPT / Claude / Gemini | 1.0×每1M输出token | 含免费模型 | 去看看 → |
| DeepInfra开源模型低价 | Llama / Qwen / DeepSeek | 按模型计费每1M输出token | 开源低价 | 去看看 → |
| Groq超低延迟 LPU | Llama / Qwen / GPT-OSS | 按模型计费每1M输出token | 高速低价 | 去看看 → |
| Fireworks AI生产级开源推理 | Llama / Qwen / DeepSeek | 按模型计费每1M输出token | 开源低价 | 去看看 → |
| 智••••资质不一 · 风险自担 | Claude / GPT / Gemini | 0.5× 起 | 仅供参考 | 注册可见 |
| 云•••••资质不一 · 风险自担 | 全模型 | 0.55× | 仅供参考 | 注册可见 |
覆盖国内官方与国外正规平台，持续更新 AI API 比价、更低价格、token价格、中转与免费额度信息。倍率以同模型官网价为基准；价格以各服务商官网实时报价为准。
我们在意您使用的服务商是否正规、可信赖，并在此基础上为您找出更低价。省钱的前提，是您的资金与数据始终安全。
按您的用法精准推送,绝不发无关的、重复的、没用的
免费订阅,每封邮件都注明您为什么收到、可一键退订。升级会员 →
先看一眼我们会发什么样的邮件 →
转发专属链接,好友点击订阅即自动记到您名下,无需填写邀请码。
Tap tokens to jump to the savings board
What you save is yours.
We track API price comparison data, free tier credits and limited-time deals across providers worldwide, including Claude API cost and discounted GPT-4 routes, then alert you the moment a better price appears. Free to subscribe.
● Live prices · compare when ready
| Provider | Models | Price | Status | |
|---|---|---|---|---|
| OpenAIGPT · Codex | GPT-5.6 / GPT-5.5 / GPT-5.4 | list priceper 1M output tokens | baseline | Visit → |
| AnthropicClaude family | Fable / Opus / Sonnet / Haiku | list priceper 1M output tokens | baseline | Visit → |
| GoogleAI Studio · Gemini | Gemini 3.1 / 3 Flash / Flash-Lite | free tierper 1M output tokens | free quota | Visit → |
| OpenRouter594 models, one API | GPT / Claude / Llama / DeepSeek | 1.0×per 1M output tokens | free models | Visit → |
| DeepSeek OfficialV4 Flash / Pro | DeepSeek V4 Flash / Pro | 1.0×per 1M output tokens | official low | Visit → |
| Groqultra-low latency LPU | Llama / Qwen / GPT-OSS | per-model pricingper 1M output tokens | fast & low | Visit → |
| Fireworks AIfast open-model serving | Llama / Qwen / DeepSeek | per-model pricingper 1M output tokens | open models | Visit → |
| DeepInfraopen-model inference | Llama / Qwen / DeepSeek | per-model pricingper 1M output tokens | open models | Visit → |
| Hyperbolicopen-weight inference | Llama / Qwen / DeepSeek | per-model pricingper 1M output tokens | open models | Visit → |
Multipliers use the same model's official list price as baseline. Some outbound links may be affiliate, sponsored, or tracked links; many are direct provider references. Prices follow each provider's live pricing page.
We only recommend providers we'd trust with our own money — reputable, reliable, and genuinely discounted. Saving money should never put your funds or your data at risk.
Tuned to how you actually use AI — never spam, never duplicates, never noise
Free to subscribe. Every email shows why you got it and has one-click unsubscribe.Membership & Pricing →
Peek at what we'll actually send you →
Share your link — friends can subscribe right from it, no codes to enter.
您的反馈我们真的会看。占用 20 秒。
We actually read these. Takes 20 seconds.