RelayModels is an OpenAI-compatible gateway that provides a single API key for accessing 39 AI models from providers including OpenAI, Anthropic, Google, DeepSeek, Qwen, xAI, Kimi, and MiniMax. Developers can switch models by changing a request parameter, while keeping their existing OpenAI SDK-based code largely unchanged. The service supports chat and text generation, document parsing, reasoning, function calling, and streaming responses. Images, audio, video, and PDFs can also be sent through the same API.

RelayModels is designed for product teams and businesses building assistants, automated replies, summarization, and other AI features. Its infrastructure handles provider failover and limits, and a dashboard provides API key management, usage controls, and request-level logs showing token usage, cost, and the responding model. Pricing is pay-as-you-go, with no subscription or minimum payment; customers top up a shared balance and pay for tokens used. The site lists pricing from $0.005 per 1,000 tokens and says balances can be topped up from 100 rubles.