Pricing

Pay per request. No subscription, no minimum spend, no charge for a failed request. Prices below are read from the gateway’s live price list.

排序
显示 77 / 77 个模型

按 token 计费

对话、embedding、语音转写 —— 每百万 tokens 的价格。

Other8

  • bge-m3向量
    输入$0.0125
  • text-embedding-3-small向量
    输入$0.025
  • mimo-v2.6-flash对话
    输入$0.175
    输出$0.35
    缓存命中$0.0035
  • mimo-v2.5对话
    输入$0.175
    输出$0.35
    缓存命中$0.0035
  • codex-auto-review对话
    输入$0.4657
    输出$2.79
    缓存命中$0.0466
  • mimo-v2.6-pro对话
    输入$0.5438
    输出$1.09
    缓存命中$0.0045
  • mimo-v2.5-pro对话
    输入$0.5438
    输出$1.09
    缓存命中$0.0045
  • MiniMax-M3对话 · 分段计费
    $1.5/ 百万 tokens

智谱4

  • glm-5.2对话
    输入$0.0563
    输出$0.1768
    缓存命中$0.0104
  • glm-5.3-flash对话
    输入$0.0926
    输出$0.3087
    缓存命中$0.0185
  • glm-5.3-flashx对话
    输入$0.4017
    输出$1.36
    缓存命中$0.0814
  • glm-5.3对话
    输入$0.9692
    输出$3.05
    缓存命中$0.18

阿里巴巴4

  • qwen3.8-flash对话
    输入$0.0926
    输出$0.2779
    缓存命中$0.009263
  • qwen3.7-plus对话
    输入$0.2792
    输出$1.12
    缓存命中$0.0558
  • qwen3.8-max对话
    输入$1.24
    输出$3.71
    缓存命中$0.1544
  • qwen3.7-max对话
    输入$1.54
    输出$4.63
    缓存命中$0.3088

DeepSeek5

  • deepseek-v4-flash对话
    输入$0.099
    输出$0.198
    缓存命中$0.00198
  • deepseek-flash对话
    输入$0.1125
    输出$0.0281
    缓存命中$0.000563
  • deepseek-v4-flash-vision-exp对话
    输入$0.2475
    输出$0.7425
    缓存命中$0.007875
  • deepseek-v3.2对话
    输入$0.252
    输出$0.378
    缓存命中$0.1215
  • deepseek-v4-pro对话
    输入$0.297
    输出$0.594
    缓存命中$0.002458

讯飞2

  • muse-spark-1.3-contributor对话
    输入$0.1245
    输出$0.249
    缓存命中$0.00249
  • muse-spark-1.3对话
    输入$1.56
    输出$5.31
    缓存命中$0.1875

Anthropic10

  • claude-haiku-4-5对话
    输入$0.1406
    输出$0.7031
    缓存命中$0.0141
  • claude-sonnet-5对话
    输入$0.2232
    输出$1.12
    缓存命中$0.0223
  • claude-opus-4-7对话
    输入$0.5581
    输出$2.79
    缓存命中$0.0558
  • claude-sonnet-4-6对话
    输入$0.6
    输出$3
    缓存命中$0.06
  • claude-sonnet-5-5对话
    输入$0.838
    输出$4.19
    缓存命中$0.0838
  • claude-opus-4-8对话
    输入$1.12
    输出$5.59
    缓存命中$0.1118
  • claude-opus-5对话
    输入$1.48
    输出$7.42
    缓存命中$0.1485
  • claude-opus-5-5对话
    输入$1.68
    输出$8.38
    缓存命中$0.0838
  • claude-fable-5-1对话
    输入$3.63
    输出$18.16
    缓存命中$0.0908
  • claude-fable-5对话
    输入$8.18
    输出$40.89
    缓存命中$0.8178

xAI3

  • grok-4.5对话
    输入$0.1489
    输出$0.4467
    缓存命中$0.0372
  • grok-4.6对话
    输入$0.1489
    输出$0.4467
    缓存命中$0.0372
  • grok-4.7对话
    输入$0.1489
    输出$0.4467
    缓存命中$0.0372

OpenAI9

  • gpt-6-luna对话 · 分段计费
    $0.25/ 百万 tokens
  • gpt-5.6-luna对话 · 分段计费
    $0.5/ 百万 tokens
  • gpt-6.1-sol对话 · 分段计费
    $5/ 百万 tokens
  • gpt-5.6-terra对话 · 分段计费
    $5/ 百万 tokens
  • gpt-6-sol对话 · 分段计费
    $5/ 百万 tokens
  • gpt-5.5对话 · 分段计费
    $12.5/ 百万 tokens
  • gpt-5.6-sol对话 · 分段计费
    $12.5/ 百万 tokens
  • gpt-6-astra对话 · 分段计费
    $25/ 百万 tokens
  • gpt-transcribe语音
    输入$93.75
    输出$187.5

Google4

  • gemini-3.8-flash对话
    输入$0.3748
    输出$1.87
    缓存命中$0.0375
  • gemini-3.7-flash对话
    输入$0.3748
    输出$1.87
    缓存命中$0.0375
  • gemini-3.6-flash对话
    输入$0.45
    输出$2.25
    缓存命中$0.045
  • gemini-3.1-pro-preview对话
    输入$0.4998
    输出$3
    缓存命中$0.05

字节跳动2

  • doubao-seed-2-1-turbo-260628对话
    输入$0.525
    输出$2.6
    缓存命中$0.105
  • doubao-seed-2-1-pro-260628对话
    输入$1.04
    输出$5.21
    缓存命中$0.2075

Moonshot4

  • kimi-k2.7-code对话
    输入$0.8906
    输出$3.75
    缓存命中$0.1781
  • kimi-k2.6对话
    输入$1.19
    输出$5
    缓存命中$0.2
  • kimi-k2.8-preview对话
    输入$1.25
    输出$5
    缓存命中$0.3125
  • kimi-k3对话
    输入$2.42
    输出$12.11
    缓存命中$0.2423

按次计费

图片与视频 —— 每次生成的价格。与上面的按 token 表不可直接比较。

字节跳动1

  • doubao-seedream-5-0-pro-260628图片
    • Image Input (after first)$0.00375
    • Image Output$0.0563
    • Layer Image Output$0.0281
    • High-Resolution Image Output$0.1125
    • +1

Other7

  • seedance-2.0-mini图片
    • 480P$0.0132
    • 480P-input$0.007909
    • 720P$0.0284
    • 720P-input$0.017
  • seedance-2.0-fast图片
    • 480P$0.0352
    • 480P-input$0.0207
    • 720P$0.0756
    • 720P-input$0.0446
  • seedance-2.0图片
    • 4K$0.9233
    • 4K-input$0.554
    • 1080P$0.4443
    • 1080P-input$0.2712
    • +4
  • MiniMax-H3图片
    • 2K$0.0975
    • 768P$0.06
  • seedance-2.5图片
    • 1080P$0.6752
    • 1080P-input$0.4039
    • 480P$0.1221
    • 480P-input$0.073
    • +2
  • midjourney-v8.2图片
    • fast$0.2064
    • relax$0.1689
    • turbo$0.375
  • midjourney-niji-7图片
    • fast$0.2064
    • relax$0.1689
    • turbo$0.375

OpenAI6

  • gpt-image-2.5-sunburst图片
    $0.009375/ 次
  • gpt-image-2图片
    $0.009375/ 次
  • gpt-image-2.5-flare图片
    $0.009375/ 次
  • gpt-image-2.5-flare-fd图片
    • 1K$0.0234
    • 2K$0.0281
    • 4K$0.0563
  • gpt-image-2.5-sunburst-fd图片
    • 1K$0.0234
    • 2K$0.0281
    • 4K$0.0563
  • gpt-image-2-fd图片
    • 1K$0.0234
    • 2K$0.0281
    • 4K$0.0563

Google4

  • gemini-3.1-flash-image图片
    • 0.5K$0.0113
    • 1K$0.0168
    • 2K$0.0253
    • 4K$0.0378
  • gemini-nano-banana-2.1图片
    • 1K$0.02
    • 2K$0.03
    • 4K$0.045
  • gemini-3-pro-image图片
    • 1K$0.0251
    • 2K$0.0251
    • 4K$0.045
  • gemini-2.5-flash-image图片
    $0.0468/ 次

xAI2

  • grok-imagine-image-2.0图片
    • 1K low$0.048
    • 1K medium$0.072
    • 2K low$0.072
    • 2K medium$0.096
    • +1
  • grok-imagine-video-1.5图片
    $0.08/ 次

快手2

  • kling-v3-omni图片
    • 4k$0.5893
    • 4k-sound$0.5893
    • base$0.0924
    • pro$0.1232
    • +4
  • kling-v3-motion-control图片
    $0.1286/ 次

This table is generated from https://api.tokwork.com/api/pricing and refreshes every five minutes. The console shows the same numbers with a per-model breakdown; text and chat models appear here as they are enabled. Switch currency in the header to see converted prices — your balance is still charged in USD, at the live rate shown in the currency menu.

How billing works

  • Prepaid balance. Top up once; every request deducts its price. When the balance reaches zero, requests stop instead of running up a bill.
  • Per-request receipts. Every call is logged with the model, the size, and the amount deducted, so you can reconcile it against your own usage.
  • Image sizes differ. For image models the price depends on the output resolution (and, for some models, the quality) — the breakdown above is the authoritative list, and it is also shown in the console before you generate.
  • Failed requests are free. A request that returns an error from a provider is not charged.

Ready to call it? Sign in to create a key, or read the quickstart first.