Skip to content

Billing

github-actions[bot] edited this page Nov 14, 2026 · 1 revision

计费说明 / Billing

计费公式 / Formula

实付价 = 标价 × 分组倍率
Price paid = List price × Group multiplier

价格单位为 美元 / 每 1M tokens,视频按秒计费的为 美元 / 秒。价格可能随时调整,请以官网实时价格为准。 Prices are in USD per 1M tokens, or USD per second for per-second video. Prices may change at any time; the live prices on the website prevail.

相当于官方价几折 / Compared with official prices

官方按美元计价,而本平台 1 元 = 1 美元额度。按 1 美元 ≈ 7.1 元估算,实际人民币成本 ≈ 官方价 × 倍率 ÷ 7.1。 Official prices are in USD, but here 1 CNY buys 1 USD of credit. Assuming 1 USD ≈ 7.1 CNY, actual cost ≈ official price × multiplier ÷ 7.1.

例如官方价 $10 的请求,在 0.3x 分组扣除 $3 额度,即 3 元人民币,约为官方价(71 元)的 4.2%。 E.g. a request costing $10 at official prices uses $3 of credit in a 0.3x group, which is 3 CNY, about 4.2% of the official 71 CNY.

国产模型本身按人民币计价,不存在汇率差:标价按 1 元 = 1 美元 直接与厂商官方人民币价对比,即「相当于官方价」= 标价 ÷ 官方价。多数模型低于官方价,见各表最后一列(官方价取自厂商价格页,2026-11)。 Chinese models are priced in CNY, so there is no exchange-rate gap: the list price is compared directly with the vendor's official CNY price at 1 CNY = 1 USD, i.e. vs. official = list price ÷ official price. Most models are below the official price; see the last column of each table (official prices from vendor pricing pages, 2026-11).

分组 / Groups

分组 Group 倍率 Multiplier 相当于官方价 vs. official pricing.json id
OpenAI 福利分组 OpenAI Welfare Group 0.06x 约 0.08 折(0.85%) openai-welfare
OpenAI 普通分组 OpenAI Standard Group 0.2x 约 0.28 折(2.8%) openai-standard
OpenAI 备用渠道 OpenAI Backup Channel 0.25x 约 0.35 折(3.5%) openai-backup
OpenAI Pro 线路 OpenAI Pro Line 0.3x 约 0.42 折(4.2%) openai-pro
OpenAI Pro 企业线路 OpenAI Pro Enterprise Line 0.5x 约 0.70 折(7.0%) openai-pro-enterprise
Claude Max 分组 Claude Max Group 1.5x 约 2.11 折(21.1%) claude-max
Claude Max 企业线路 Claude Max Enterprise Line 1.9x 约 2.68 折(26.8%) claude-max-enterprise
Gemini Chat / Coding Gemini Chat / Coding 0.3x 约 0.42 折(4.2%) gemini-chat-coding
国模分组 - Deepseek / Glm(含 Kimi / 豆包) CN Models – DeepSeek / GLM (incl. Kimi / Doubao) 1x 约 5–10 折(因模型而异) cn-models
视频生成分组 - seedance/h3 Video Generation – seedance/h3 1x 约 1.41 折(14.1%) video

如何选择分组 / Choosing a group

  • 倍率越低越便宜。标注「限时」的分组价格可能调整或下线。
    Lower multiplier = cheaper. Groups tagged limited-time may change or be withdrawn.
  • 企业线路(OpenAI Pro 企业线路、Claude Max 企业线路)为专属分组,提供 SLA 保障,适合生产环境。
    Enterprise lines (OpenAI Pro Enterprise, Claude Max Enterprise) are dedicated and come with an SLA; use them in production.
  • 同一模型可在多个分组中使用;可以为不同用途创建不同分组的 Key。
    The same model is available in several groups; create separate keys per group for different workloads.

计费细节 / Details

项目 Item 说明 Notes
缓存读取 / Cache read 命中缓存的输入 token 按「缓存读取」价计费,通常远低于输入价 / Cached input tokens are billed at the cache-read price, usually far below input
缓存写入 / Cache write 写入缓存的 token 按「缓存写入」价计费;1 小时缓存写入:claude-opus-5-5 为 $12.00,claude-haiku-5-5 为 $0.30。 / Tokens written to cache use the cache-write price; 1-hour cache write: claude-opus-5-5 $12.00, claude-haiku-5-5 $0.30.
长上下文 / Long context 部分 OpenAI 模型超过 272K 上下文后单价更高 / Some OpenAI models cost more above 272K context
GLM 输入 <32K / 输入 ≥32K 价格不同 / Input <32K and Input ≥32K are priced differently
DeepSeek 工作日高峰时段 09:00–12:00、14:00–18:00(GMT+8)价格 ×2 / Weekday peak hours 09:00–12:00 and 14:00–18:00 (GMT+8): price ×2
视频 / Video 按秒(美元 / 秒)或按 token 计费,价格随分辨率变化 / Billed per second (USD/s) or per token, varying by resolution

各模型价格见 模型与价格 / Models and Pricing,最终以官网实时价格为准。 See 模型与价格 / Models and Pricing for every price. Live prices on the website prevail.