# SiliconFlow

> AI cloud platform providing high-speed inference for LLMs, image, video, and audio models with serverless, fine-tuning, and reserved GPU options.

SiliconFlow is a comprehensive AI cloud platform that delivers high-performance inference for text, image, video, and audio models through a single unified API. The platform supports both open-source and commercial models from providers like DeepSeek, Qwen, OpenAI, MiniMax, and more, enabling developers to build AI-powered applications with predictable costs and blazing-fast performance.

- **Serverless Inference** allows running any model instantly with no setup required—just one API call with pay-per-use pricing and $1 in free credits to get started.
- **Fine-tuning Capabilities** enable customizing powerful models to specific use cases with one-click deployment for tailored AI solutions.
- **Reserved GPUs** provide guaranteed GPU capacity with NVIDIA H100/H200, AMD MI300, and RTX 4090 for stable performance and predictable billing.
- **Elastic GPUs** offer flexible FaaS deployment with reliable and scalable inference for variable workloads.
- **AI Gateway** delivers unified access with smart routing, rate limits, and cost control across all models.
- **OpenAI-Compatible API** ensures seamless integration with existing workflows through a fully compatible interface.
- **Multi-Modal Support** covers LLMs, image generation (FLUX, Z-Image), video generation (Wan2.1/2.2), and audio models (Fish-Speech, CosyVoice).
- **Privacy-First Architecture** ensures no data is stored—models and data remain under user control.
- **Transparent Pricing** with per-token billing for chat models and per-output pricing for media generation, with no hidden fees or commitments.

To get started, sign up for an account at cloud.siliconflow.com, obtain an API key, and begin making API calls immediately. The platform provides comprehensive documentation and code examples for quick integration. SiliconFlow also maintains open-source projects including OneDiff (a lightning-fast inference engine for diffusion models) and BizyAir (an AI-native runtime for scalable inference workloads).

## Features
- Serverless model inference
- Model fine-tuning
- Reserved GPU capacity
- Elastic GPU deployment
- AI Gateway with smart routing
- OpenAI-compatible API
- LLM inference
- Image generation
- Video generation
- Audio processing and synthesis
- Multi-model support
- Pay-per-use pricing
- Spending limits control
- Volume discounts
- No data storage policy

## Integrations
DeepSeek models, Qwen models, OpenAI models, MiniMax models, Moonshot AI models, Zhipu AI models, FLUX image models, Wan video models, Fish-Speech audio, CosyVoice audio

## Platforms
WEB, API

## Pricing
Freemium — Free tier available with paid upgrades

## Links
- Website: https://www.siliconflow.com
- Documentation: https://docs.siliconflow.com/
- Repository: https://github.com/siliconflow
- EveryDev.ai: https://www.everydev.ai/tools/siliconflow