DeepSeek-V4.1-Flash Now Available · 1M context, web search, and tool calling fully supportedPricing Overview
Transparent pricing, pay-as-you-go. Aggregating mainstream model providers for one-stop cost comparison, helping you quickly select the best value plan.
Serverless Inference
Pay-as-you-go model inference via unified OpenAI-compatible API, with automatic elastic scaling.
| Model | Input Price ($/M Tokens) | Output Price ($/M Tokens) | Cache Hit ($/M Tokens) |
|---|---|---|---|
 DeepSeek-v4-Pro deepseek-v4-pro | ¥3 | ¥6 | ¥0.025 |
 GLM-5.1 glm-5.1 | ¥6 | ¥28 | ¥1.3 |
GLM-5.2 glm-5.2 | ¥8 | ¥28 | ¥2 |
 Qwen3.7-Max qwen3.7-max | ¥6 | ¥18 | - |
Qwen3.7-Plus qwen3.7-plus | ¥1.6 | ¥6.4 | - |
 Kimi-k2.7-Code kimi-k2.7-code | ¥6.5 | ¥27 | - |
| Tool | Price |
|---|---|
Web search | ¥0.03 |
Dedicated Cluster
Physically isolated, exclusive compute instances. Resources are dedicated and not shared. Quoted by instance spec and region.
Quoted by Instance Spec
Dedicated clusters are billed by GPU instance. Pricing varies by hardware spec and region. Annual/monthly subscriptions available.
Learn about Dedicated Cluster →Compute Hosting
Deploy MaaS software to your own data center. Idle compute resources are integrated into the platform for unified scheduling and management. Quoted by managed scale and service tier.
Solution in Development
Compute hosting is priced by managed node scale and service tier. Contact us for a solution and quote.
Learn about Compute Hosting →Model Production
Full-lifecycle model training, fine-tuning, and deployment services. Supports pre-training, continuous pre-training, SFT/DPO fine-tuning. Quoted on demand.
Solution in Development
Contact us for model production solutions and pricing.
Learn about Model Production →