RouterHub

RouterHub is a prepaid, usage-based AI API gateway that connects applications and coding tools to models from OpenAI, Anthropic, Google, xAI, DeepSeek, Z.ai and Qwen through a single API key. The associated GitHub repository documents the gateway, including uncensored and abliterated model variants where available, with setup guides and quickstarts.

What It Is

RouterHub is an API gateway rather than a model owner. A user creates an account, adds prepaid credits, creates an API key and sends requests to a RouterHub base URL. Each model supports Chat Completions, Responses or Messages API formats, and the catalogue lists the exact model IDs and supported protocols. The site's homepage lists 35 live models.

How the Workflow Runs

Requests deduct usage charges from the prepaid balance, and balances can be topped up by card or crypto. The docs show a curl example against the Responses endpoint, and the README provides Python examples using the OpenAI and Anthropic SDKs. Separate API keys can isolate projects and usage, and prompt caching is described as lowering the cost of repeated input.

Client and Tool Compatibility

The docs include setup guides for OpenCode, VS Code with Continue, VS Code with Cline, Claude Code and Codex. The README also lists Cursor, LangChain and LlamaIndex via OpenAI-compatible adapters, and any client that accepts a custom base URL.

Uncensored Variants and Limits

The README says uncensored and abliterated variants are offered alongside standard models. It states that these reduce refusal behavior but do not guarantee that a model will never refuse. Behavior depends on the model, version, prompt and provider policy, and users remain responsible for lawful use and compliance with provider terms.