AtmoRouter

About AtmoRouter

AtmoRouter is a prepaid API gateway for AI inference that consolidates access to 171 frontier models from multiple providers including Claude, GPT, Gemini, and DeepSeek. Users connect through a single OpenAI-compatible endpoint with one API key and pay only for tokens consumed, with no monthly subscription or minimum spend required.

The service offers automatic failover across upstream capacity, ensuring requests land on healthy providers. Billing is exact per-token metering from upstream usage reports, never estimated from character counts. Streaming is fully supported with SSE passthrough identical to OpenAI's format, and users can cancel mid-generation to stop paying immediately.

Key features include unlimited API keys per account with per-key rate limits and model allowlists, complete usage logs with cost breakdowns by model and family, and transparent pricing that shows savings compared to official provider rates. The platform supports both OpenAI Chat Completions and Anthropic Messages API protocols.

• failover
• llm routing
• cost optimization
• per-token billing
• ai inference
• openai compatible
• api gateway

FAQ

What is AtmoRouter?

AtmoRouter is a prepaid API gateway for AI inference that provides a single account and API key to access models in the catalogue, with usage charged to your prepaid balance instead of requiring a monthly subscription.

How is usage charged?

Usage is charged from your prepaid balance based on recorded input, cached input, and output tokens using the applicable request-time billing rules. Billing is computed from the upstream usage report, not estimated from characters.

How do I add funds to my account?

Use the dashboard's top-up flow with payment methods shown there, including local QR payments and supported stablecoin transfers when available. Follow the invoice or deposit instructions exactly, and only a confirmed, credited payment increases your spendable balance.

Are failed or interrupted requests charged?

A failure before output is not charged. A partial or interrupted response can be billed for processed usage. Use the recorded charge and receipt in Usage rather than assuming every error or cancelled stream costs zero.

Does AtmoRouter train models on my API requests?

No, AtmoRouter does not use API request content to train models. The infrastructure serving the selected model receives the content needed to run inference, and its handling can differ by provider.