SiliconFlow
Pay-as-you-go hosted inference
SiliconFlow billing summary
Platform pricing
Published USD model rates
Routing
OpenAI-compatible API
Access
Text and multimodal models
Pricing source checked Sep 18, 2026. Official pricing
SiliconFlow pricing and free credits
Fee rules checked .
- Is SiliconFlow free?
- New international accounts can start with $1 in free credits. This is a limited balance for trying the service, not unlimited free API usage. Paid and free model variants can also have different rate limits. Source
- How does usage billing work?
- There is no minimum commitment for on-demand usage. Chat models charge separately for input and output tokens. Image, video and audio models use task-specific units, so compare the unit shown for the exact model. Source
- Can I limit spending?
- The international account dashboard supports monthly spending limits. Set a limit before production traffic if you need a billing guardrail; it does not change the model's published unit price. Source
- Are volume discounts automatic?
- SiliconFlow asks high-usage customers to contact sales for a custom pricing plan. The public page does not publish one universal volume-discount percentage. Source
How SiliconFlow compares
Compare all LLM gatewaysBYOK
No. SiliconFlow hosts and bills the models in its catalog
Routing
One API for the models published in the SiliconFlow catalog
Observability
Usage and billing records are available in the platform account
Deployment and data controls
Managed inference, reserved instances and private deployment options
Country and region availability
Sources checked 2026-08-24The displayed catalog and USD prices come from SiliconFlow's international service.
- Country support status
- The international service is collected separately from the mainland China service and does not inherit its catalog.
- Access limits
- Model availability and account eligibility can differ between the China and international services.
- Regions and data location
- The public pricing page does not establish the processing location for every model request.
Check the gateway terms and the selected model provider before relying on access in a specific country. Network edge locations, billing currencies, and provider catalogs are not country-availability evidence.
Model catalog
28 models in this platform's catalog
Model vendor
Cached input price
Minimum context
Token price order
| Model | Input / 1M | Output / 1M | Cached input | Context |
|---|---|---|---|---|
 DeepSeek-V4-Flashdeepseek-aiSource and checked dateChecked Official quote sourceOriginal / 1M tokens: USD 0.13 input, 0.28 output | $0.13 | $0.28 | $0.028 | 1.1M |
DeepSeek-V4-Flash-0731deepseek-aiSource and checked dateChecked Official quote sourceOriginal / 1M tokens: USD 0.22 input, 0.66 output | $0.22 | $0.66 | $0.014 | 1.1M |
DeepSeek-V4-Flash-Vision-Expdeepseek-aiSource and checked dateChecked Official quote sourceOriginal / 1M tokens: USD 0.44 input, 1.32 output | $0.44 | $1.32 | $0.028 | 1.1M |
DeepSeek-V4-Prodeepseek-aiSource and checked dateChecked Official quote sourceOriginal / 1M tokens: USD 1.50162 input, 3.135 output | $1.50162 | $3.135 | $0.135 | 1.1M |
DeepSeek-V4-Pro-0813deepseek-aiSource and checked dateChecked Official quote sourceOriginal / 1M tokens: USD 1.32 input, 3.96 output | $1.32 | $3.96 | $0.044 | 1.1M |
 gemma-4-12B-itgoogle | $0.1 | $0.3 | Not listed | 268K |
gemma-4-26B-A4B-itgoogleSource and checked dateChecked Official quote sourceOriginal / 1M tokens: USD 0.12 input, 0.4 output | $0.12 | $0.4 | Not listed | 268K |
gemma-4-31B-itgoogleSource and checked dateChecked Official quote sourceOriginal / 1M tokens: USD 0.13 input, 0.4 output | $0.13 | $0.4 | Not listed | 268K |
| MiniMax-M2.5MiniMaxAI | $0.3 | $1.2 | $0.03 | 202K |
| MiniMax-M3MiniMaxAI | $0.3 | $1.2 | $0.06 | 1.1M |
Model rates come from the platform's published model catalog. Token rates exclude taxes, payment charges, negotiated discounts, and separately published plans.