Pricing

Models and prices

Every model on sale, in US dollars per million tokens. You pay per token used, from prepaid credit bought under an offline contract.

ModelContextInput / 1M tokensOutput / 1M tokensPrompt cache and service tiers
DeepSeek-V4-Prodeepseek-ai/deepseek-v4-proPrice version tencent-cn-20261119-c18c1e4aUpdated 1M tokens$1.7876296$3.57525921
  • Prompt cache: $0.14896913 read / $1.7876296 write
DeepSeek-V4-Pro 0813 正式版deepseek-ai/deepseek-v4-pro-0813Price version deepseek-official-6a7c258e-5Updated 1.02M tokens$0.66$1.98
  • Prompt cache: $0.022 read / $0.66 write
DeepSeek-V4.1-Flashdeepseek-ai/deepseek-v4.1-flashPrice version deepseek-official-0d635320-5Updated 1M tokens$0.15$0.60
  • Prompt cache: $0.003 read / $0.15 write
DeepSeek-V4.1-Flash-1017deepseek-ai/deepseek-v4.1-flash-1017Price version tencent-cn-20261119-c18c1e4a-20261120T1000Updated 1.02M tokens$0.14896913$0.59587653
  • Prompt cache: $0.00297938 read / $0.14896913 write
MiniMax-M2.5minimax/minimax-m2.5Price version qwen-cn-20261119-93663bb9Updated 200K tokens$0.31283518$1.25134072
  • Prompt cache: $0.06256704 read / $0.31283518 write
MiniMax-M3minimax/minimax-m3Long prompts cost more: over 512K prompt tokens, $0.62567036 in / $2.50268144 out, per 1M tokens.Price version tencent-cn-20261119-c18c1e4aUpdated 1M tokensfrom $0.31283518from $1.25134072
  • Prompt cache: $0.06256704 read / $0.31283518 write
MiniMax-M3.1-Flashminimax/minimax-m3.1-flashPrice version tencent-cn-20261119-c18c1e4aUpdated 1.02M tokens$0.14896913$0.59587653
  • Prompt cache: $0.01489691 read / $0.14896913 write
Kimi K2.6moonshotai/kimi-k2.6Price version tencent-cn-20261119-c18c1e4aUpdated 256K tokens$0.96829937$4.02216661
  • Prompt cache: $0.16386605 read / $0.96829937 write
Kimi K2.7 Codemoonshotai/kimi-k2.7-codePrice version tencent-cn-20261119-c18c1e4aUpdated 256K tokens$0.96829937$4.02216661
  • Prompt cache: $0.19365987 read / $0.96829937 write
Kimi K2.7 Code HighSpeedmoonshotai/kimi-k2.7-code-highspeedPrice version tencent-cn-20261119-c18c1e4aUpdated 256K tokens$1.93659874$8.04433321
  • Prompt cache: $0.38731975 read / $1.93659874 write
Kimi K2.8 Previewmoonshotai/kimi-k2.8-previewPrice version tencent-cn-20261119-c18c1e4aUpdated 1.02M tokens$0.96829937$4.02216661
  • Prompt cache: $0.25324753 read / $0.96829937 write
Kimi K3moonshotai/kimi-k3Price version tencent-cn-20261119-c18c1e4aUpdated 1M tokens$2.97938267$14.89691336
  • Prompt cache: $0.29793827 read / $2.97938267 write
Kimi K3.1 Promoonshotai/kimi-k3.1-proPrice version qwen-cn-20261119-93663bb9Updated 1M tokens$1.93659874$10.12990108
  • Prompt cache: $0.19365987 read / $1.93659874 write
Kimi K3.1 Pro HighSpeedmoonshotai/kimi-k3.1-pro-highspeedPrice version tencent-cn-20261119-c18c1e4aUpdated 1.02M tokens$3.87319747$20.25980217
  • Prompt cache: $0.19365987 read / $3.87319747 write
GUI-Plusqwen/gui-plusPrice version qwen-cn-20261119-93663bb9Updated 256K tokens$0.2234537$0.6703611—
Qwen-Flash-Characterqwen/qwen-flash-characterPrice version qwen-cn-20261119-93663bb9Updated 8K tokens$0.03724228$0.2234537
  • Prompt cache: $0.00744846 read / $0.03724228 write
Qwen-MT-Flashqwen/qwen-mt-flashPrice version qwen-cn-20261119-93663bb9Updated 16K tokens$0.10427839$0.29048981—
Qwen-MT-liteqwen/qwen-mt-litePrice version qwen-cn-20261119-93663bb9Updated 16K tokens$0.08938148$0.23835061—
Qwen-MT-Plusqwen/qwen-mt-plusPrice version qwen-cn-20261119-93663bb9Updated 16K tokens$0.26814444$0.80443332—
Qwen-Plus-Characterqwen/qwen-plus-characterPrice version qwen-cn-20261119-93663bb9Updated 32K tokens$0.11917531$0.29793827
  • Prompt cache: $0.02383506 read / $0.14896913 write
Qwen-VL-OCRqwen/qwen-vl-ocrPrice version qwen-cn-20261119-93663bb9Updated 38.19K tokens$0.04469074$0.07448457—
Qwen3-VL-Plusqwen/qwen3-vl-plusLong prompts cost more: over 32K prompt tokens, $0.2234537 in / $2.234537 out; over 128K prompt tokens, $0.4469074 in / $4.46907401 out, per 1M tokens.Price version qwen-cn-20261119-93663bb9Updated 256K tokensfrom $0.14896913from $1.48969134
  • Prompt cache: $0.02979383 read / $0.18621142 write
Qwen3.5-OCRqwen/qwen3.5-ocrPrice version qwen-cn-20261119-93663bb9Updated 64K tokens$0.07448457$0.29793827—
Qwen3.5-Omni-Plusqwen/qwen3.5-omni-plusPrice version qwen-cn-20261119-93663bb9Updated 256K tokens$1.04278394$5.95876534—
Qwen3.7-Flashqwen/qwen3.7-flashLong prompts cost more: over 32K prompt tokens, $0.08938148 in / $0.35752592 out; over 256K prompt tokens, $0.17876296 in / $0.71505184 out, per 1M tokens.Price version qwen-cn-20261119-93663bb9Updated 1M tokensfrom $0.02979383from $0.11917531
  • Prompt cache: $0.00595877 read / $0.03724228 write
Qwen3.7-Maxqwen/qwen3.7-maxPrice version qwen-cn-20261119-93663bb9Updated 1M tokens$1.7876296$5.36288881
  • Prompt cache: $0.35752592 read / $2.234537 write
Qwen3.7-Plusqwen/qwen3.7-plusLong prompts cost more: over 256K prompt tokens, $0.71505184 in / $2.86020737 out, per 1M tokens.Price version qwen-cn-20261119-93663bb9Updated 1M tokensfrom $0.23835061from $0.95340246
  • Prompt cache: $0.04767012 read / $0.29793827 write
Qwen3.7-Text-Embedding-Flashqwen/qwen3.7-text-embedding-flashEmbeddingsPrice version qwen-cn-20261119-93663bb9Updated 128K tokens$0.01862114$0.00—
Qwen3.8-2.4T-A95Bqwen/qwen3.8-2.4t-a95bPrice version qwen-cn-20261119-93663bb9Updated 1M tokens$1.7876296$5.36288881
  • Prompt cache: $0.2234537 read / $2.234537 write
Qwen3.8-Flashqwen/qwen3.8-flashPrice version qwen-cn-20261119-93663bb9Updated 1M tokens$0.11917531$0.40221666
  • Prompt cache: $0.01489691 read / $0.18621142 write
Qwen3.8-Maxqwen/qwen3.8-maxPrice version qwen-cn-20261119-93663bb9Updated 1M tokens$1.7876296$5.36288881
  • Prompt cache: $0.2234537 read / $2.234537 write
Qwen3.8-Omni-Flashqwen/qwen3.8-omni-flashPrice version qwen-cn-20261119-93663bb9Updated 1M tokens$0.11917531$0.40221666
  • Prompt cache: $0.01489691 read / $0.11917531 write
Qwen4-Maxqwen/qwen4-maxPrice version qwen-cn-20261119-93663bb9Updated 1M tokens$1.48969134$4.46907401
  • Prompt cache: $0.07448457 read / $1.86211417 write
伶鹊-对话分析-Flashqwen/tongyi-xiaomi-analysis-flashPrice version qwen-cn-20261119-93663bb9Updated 32K tokens$0.02979383$0.05958765
  • Prompt cache: $0.00595877 read / $0.02979383 write
伶鹊-对话分析-Proqwen/tongyi-xiaomi-analysis-proPrice version qwen-cn-20261119-93663bb9Updated 32K tokens$0.14896913$0.40221666
  • Prompt cache: $0.02979383 read / $0.14896913 write
Step 5 Previewstepfun/step-5-previewPrice version tencent-cn-20261119-c18c1e4aUpdated 1M tokens$1.04278394$2.97938267
  • Prompt cache: $0.0521392 read / $1.04278394 write
Hy-MT2-Litetencent/hy-mt2-litePrice version tencent-cn-20261119-c18c1e4aUpdated 8K tokens$0.04469074$0.17876296—
Hy-MT2-Plustencent/hy-mt2-plusPrice version tencent-cn-20261119-c18c1e4aUpdated 8K tokens$0.07448457$0.29793827—
Hy-MT2-Protencent/hy-mt2-proPrice version tencent-cn-20261119-c18c1e4aUpdated 8K tokens$0.07448457$0.29793827—
Hy-Roletencent/hy-rolePrice version tencent-cn-20261119-c18c1e4aUpdated 32K tokens$0.35752592$1.43010368—
Hy-Role-Latesttencent/hy-role-latestPrice version tencent-cn-20261119-c18c1e4aUpdated 32K tokens$0.35752592$1.43010368—
Hy3tencent/hy3Price version tencent-cn-20261119-c18c1e4aUpdated 256K tokens$0.14896913$0.59587653
  • Prompt cache: $0.03724228 read / $0.14896913 write
Hy4tencent/hy4Price version tencent-cn-20261119-c18c1e4aUpdated 1.02M tokens$0.4469074$1.3407222
  • Prompt cache: $0.02234537 read / $0.4469074 write
Hy4 flashtencent/hy4-flashPrice version tencent-cn-20261119-c18c1e4aUpdated 1.02M tokens$0.03724228$0.14896913
  • Prompt cache: $0.00148969 read / $0.03724228 write
MiMo-V2.6-Flashxiaomi/mimo-v2.6-flashPrice version tencent-cn-20261119-c18c1e4aUpdated 1.02M tokens$0.14896913$0.29793827
  • Prompt cache: $0.00297938 read / $0.14896913 write
MiMo-V2.6-Proxiaomi/mimo-v2.6-proPrice version tencent-cn-20261119-c18c1e4aUpdated 1.02M tokens$0.4469074$0.8938148
  • Prompt cache: $0.00372423 read / $0.4469074 write
GLM-4-Flash-250414zai-org/glm-4-flash-250414Price version zhipu-standard-20261119Updated 128K tokens$0.00$0.00—
GLM-5zai-org/glm-5Long prompts cost more: from 32K prompt tokens, $0.8938148 in / $3.27732094 out, per 1M tokens.Price version zhipu-standard-20261119Updated 200K tokensfrom $0.59587653from $2.6814444
  • Prompt cache: $0.14896913 read / $0.59587653 write
GLM-5-Turbozai-org/glm-5-turboLong prompts cost more: from 32K prompt tokens, $1.04278394 in / $3.87319747 out, per 1M tokens.Price version zhipu-standard-20261119Updated 200K tokensfrom $0.74484567from $3.27732094
  • Prompt cache: $0.17876296 read / $0.74484567 write
GLM-5.1zai-org/glm-5.1Long prompts cost more: from 32K prompt tokens, $1.19175307 in / $4.17113574 out, per 1M tokens.Price version zhipu-standard-20261119Updated 200K tokensfrom $0.8938148from $3.57525921
  • Prompt cache: $0.19365987 read / $0.8938148 write
GLM-5.2zai-org/glm-5.2Price version zhipu-standard-20261119Updated 1M tokens$1.19175307$4.17113574
  • Prompt cache: $0.29793827 read / $1.19175307 write
GLM-5.3zai-org/glm-5.3Price version zhipu-standard-20261119Updated 1.02M tokens$1.19175307$4.17113574
  • Prompt cache: $0.29793827 read / $1.19175307 write
GLM-5.3-Flashzai-org/glm-5.3-flashPrice version zhipu-standard-20261119Updated 800K tokens$0.11917531$0.41711357
  • Prompt cache: $0.0342629 read / $0.11917531 write
GLM-5.3-FlashXzai-org/glm-5.3-flashxPrice version zhipu-standard-20261119Updated 1.02M tokens$0.29793827$1.04278394
  • Prompt cache: $0.08491241 read / $0.29793827 write

About these prices

  • You pay for the tokens you send at the input price, and the tokens the model writes at the output price — a reasoning model’s reasoning included.
  • These are Token Gear’s prices, the ones in effect now. Signed in, the console’s Models page shows the prices your account pays.
  • A price marked “from” is the lowest: a long prompt costs more, as its row says. Where the prompt cache is priced apart, its row says what the part of a prompt read from or written to the cache costs. Your account’s data constraints keep requests away from providers that do not meet them (Privacy Policy §4).
  • Prices exclude taxes unless your contract says otherwise (Terms §4). Credit is added by an administrator under an offline contract: there is no self-service top-up or card payment (contact sales). How retries and streams cut short are billed is in the API docs.