拿到 Key，就能用
Take a key, start using it
无注册，不记名，拿到就能用 No sign-up, no name attached, just a key
没有账户体系，没有订阅，没有绑卡。一个 Key 就是全部凭证——改三行配置立刻开始，用多少扣多少。 No account system, no subscription, no card on file. The key is the whole credential — change three lines, start now, pay only for what you use.
0
账户 / 实名 / 订阅费accounts, ID checks, subscriptions
3
行代码完成接入lines of code to integrate
1
不同模型独立定价independent rate per model
$3¥19
最低每 1M tokenslowest per 1M tokens
两种卡Two kinds of key
按 token 买，或者按金额买Buy tokens, or buy an amount
Token 卡Token key
指定模型 + 指定 token 数One model, one token total
「这张卡有 2000 万 token」就是全部说明书。“20M tokens” is the entire manual.
价格 = 统一价 × token 数（M）price = flat rate × tokens in millions
- Kimi-K2.7-Code · 20M → $60¥380Kimi-K2.7-Code · 20M → $60¥380
- GPT-5.6 luna · 20M → $120¥780GPT-5.6 luna · 20M → $120¥780
- GPT-5.6 sol · 5M → $150¥995GPT-5.6 sol · 5M → $150¥995
金额卡Amount key
一张卡，多个模型One key, several models
余额按各模型的统一价实时扣减，还没想好用哪个模型时先拿它跑。A balance drawn down at each model's flat rate — the one to take while you are still choosing.
可用 token = 金额 ÷ 统一价 × 1Mtokens = amount ÷ flat rate × 1M
- $10¥99 → Kimi 约 3.3M5.2M tokens$10¥99 → about 3.3M5.2M Kimi tokens
- $50¥499 → Kimi 约 16.7M26.3M tokens$50¥499 → about 16.7M26.3M Kimi tokens
- $200¥1,999 → Kimi 约 66.7M105M tokens$200¥1,999 → about 66.7M105M Kimi tokens
怎么选：馈赠、商务礼品、活动发给学员、做定额试用 → Token 卡（数字直观、可印在卡片上）。自用、团队用、要在多个模型之间切换 → 金额卡。
Which one: gifts, corporate presents, handouts at an event, a fixed trial → token key (a number that reads well on a card). Your own use, a team, switching between models → amount key.
价目表Rate card
每 1M tokens 统一价Flat rate per 1M tokens
| 模型Model | 适合什么Good for | 每 1M tokensPer 1M tokens |
|---|---|---|
| GPT-5.6 sol旗舰Flagship | 同代最强，指令遵循和工具调用更稳Strongest of its generation; steadier instruction following and tool use | $30¥199 |
| GPT-5.6 terra均衡Balanced | 日常主力，能力和价格的平衡点The daily driver — the balance point of capability and cost | $15¥99 |
| GPT-5.6 luna均衡Balanced | 高频调用、对话与轻量 AgentHigh-frequency calls, chat and lightweight agents | $6¥39 |
| DeepSeek-V4-Pro低价Economy | 批量任务、数据处理、大规模抽取Batch jobs, data processing, large-scale extraction | $4¥29 |
| Kimi-K2.7-Code低价Economy | 编码特化，Agent 长会话最省的一档Code-specialised; the cheapest option for long agent runs | $3¥19 |
场景预估Consumption estimates
你的用法，大概烧多少 tokenRoughly how much your workload burns
| 场景Scenario | 单次约Per request | 输入 / 输出占比Input / output split | 1M tokens 约可跑1M tokens buys |
|---|---|---|---|
| 对话问答Chat Q&A | 1,300 |
62% / 38%
|
约 770 次~770 calls |
| 批量生成Bulk generation | 4,800 |
38% / 62%
|
约 208 次~208 calls |
| RAG 检索问答RAG retrieval | 8,400 |
95% / 5%
|
约 119 次~119 calls |
| Agent 长会话Long agent runs | 17,500 |
86% / 14%
|
约 57 次~57 calls |
| 编码助手Coding assistant | 25,500 |
94% / 6%
|
约 39 次~39 calls |
| 长文档分析Long-doc analysis | 42,000 |
95% / 5%
|
约 24 次~24 calls |
输入（含被缓存复用的上下文）Input (including cached context)
输出（模型生成的部分）Output (what the model writes)