Leading Chinese LLM inference cloud service
Introduction
SiliconFlow is a leading Chinese LLM inference cloud hosting mainstream open models (DeepSeek, Qwen, GLM and more) in one place, with inference price-performance and domestic access speed as core edges. Free tiers cover several small models — a popular on-ramp for indie devs and startups to flagship open models at low cost.
Key Features
- One-stop hosting of mainstream open models
- Optimized inference price-performance
- Free small-model allowances
- OpenAI-compatible API
- Private and enterprise plans
Pros
- Dual edge in domestic speed and price
- Fast model lineup updates
Cons
- Flagship commercial models rely on partnerships
Pricing
Per-token billing with free models and granted credits.（See official site for current pricing）
FAQ
DeepSeek pricing?
Per-token, often below official direct pricing tiers.
Stable?
Production-grade; occasional queuing at peaks.
Fine-tuning?
Yes, fine-tuning services are offered.
Related Sites
Collected on lucifer.city · Reviewed by the founder