SiliconFlow BYOK Setup Guide: Use One Key in Cherry Studio, ChatHub, Claude Code & More

BYOK stands for Bring Your Own Key. It means a third-party AI tool lets you enter your own API key and choose your own model backend, rather than forcing you to use the tool creator's servers. The tool sends requests to whatever endpoint you specify.

SiliconFlow is ideal for BYOK because its API is fully OpenAI-compatible. Any tool that accepts a custom OpenAI base URL will work with SiliconFlow out of the box — no special integration needed. You get access to 200+ models in your favorite tool without running anything locally. The complete SiliconFlow resource covers the full range of capabilities beyond BYOK.

This is especially powerful for free-tier users: you can use the free Qwen3-8B model inside premium chat clients and translation tools at zero cost. If you haven't obtained a SiliconFlow API key yet, registration takes just a few minutes.

Cherry Studio is a popular open-source desktop client that lets you chat with multiple AI models side by side. It natively supports custom OpenAI-compatible providers, making SiliconFlow integration seamless.

To set it up: open Cherry Studio Settings, navigate to Model Providers, click Add Provider, and fill in the fields. Set the provider name to SiliconFlow, the API endpoint to https://api.siliconflow.cn/v1, and paste your API key. Then add the model IDs you want to use (e.g., Qwen/Qwen3-8B, deepseek-ai/DeepSeek-V4-Pro) in the models field.

ChatHub is a browser extension (available for Chrome, Edge, and Firefox) that lets you compare responses from multiple AI models simultaneously. It supports custom OpenAI-compatible endpoints, so SiliconFlow works directly.

In ChatHub settings, go to Custom AI Providers, add a new provider, and enter the same endpoint and key as above. The advantage of ChatHub is that you can see responses from Qwen3, DeepSeek, and GLM side by side in real time, helping you pick the best model for each query.

Immersive Translate is a popular translation tool that can use AI models for higher-quality translations. By default it requires its own subscription, but it also supports BYOK for OpenAI-compatible APIs. Configuring SiliconFlow as the backend gives you access to powerful Chinese-optimized models at a fraction of the cost.

In Immersive Translate settings, find the AI Translation service section, select Custom OpenAI-compatible, and set the base URL to https://api.siliconflow.cn/v1. Use a model like Qwen/Qwen3-8B (free) or deepseek-ai/DeepSeek-V4-Pro (paid) depending on your quality and budget needs.

Claude Code is an AI coding agent that normally uses Anthropic's Claude models. However, because it relies on an OpenAI-compatible API layer, you can point it at SiliconFlow and use DeepSeek-V4-Pro or Qwen3 as the coding backend instead — often at much lower cost.

Configuration involves setting environment variables. Set ANTHROPIC_BASE_URL (or the relevant provider URL variable) to https://api.siliconflow.cn/v1 and your SiliconFlow key as the API key. Then specify the model ID in your Claude Code configuration. DeepSeek-V4-Pro is the recommended model for coding tasks due to its strong reasoning and code-generation benchmarks.

For any other tool that supports custom OpenAI endpoints, the configuration is always the same three values. Once you know this pattern, you can connect SiliconFlow to virtually any BYOK-compatible tool.

If a tool asks for an API key prefix, SiliconFlow keys start with sk-. If a tool requires a model list, use the full vendor/model-name format. For guidance on choosing the right model for your use case and budget, see our model comparison guide. If a tool has a connection test feature, test with a free model first (Qwen/Qwen3-8B) to confirm without spending credits.