One OpenAI-compatible API for GPT, Claude, Gemini and image models.One endpoint.
Build against one endpoint, switch models instantly, and manage usage, billing, API keys, and logs from a single dashboard.
$ curl https://api.novagate.ai/v1/chat/completions \
-H "Authorization: Bearer $NOVAGATE_KEY" \
-d '{"model":"gpt-5","messages":[{"role":"user","content":"Hi"}]}'Same models. Up to 20% off.
We negotiate volume rates with every major lab and pass the savings to you. Average 20% cheaper than going direct — same weights, same outputs, lower bill.
Lower bill. Higher uptime.
Swap one base URL — keep your SDK, keep your code, cut the cost.
Up to 50% cheaper
Volume rates with every major lab, passed straight to you. Transparent per-token pricing — no surprise invoices.
99.99% uptime SLA
Multi-region failover and automatic provider rerouting. If one lab is degraded, your requests still complete.
Every major model
GPT-5, Claude, Gemini, DeepSeek, Llama, Mistral — added the day they launch. One unified API, one bill.
Pay as you go.
Top up credits. They never expire.
For active developers shipping real apps.
Start with BuilderQuestions, answered.
What is Nova AI Relay?+
Nova AI Relay is an OpenAI-compatible AI gateway that lets developers access GPT, Claude, Gemini, image models, and other model families through one account, one API key, and one billing dashboard.
Is Nova AI Relay compatible with the OpenAI API?+
Yes. Nova AI Relay is designed around OpenAI-compatible endpoints, so many OpenAI SDKs, CLI tools, and agent frameworks can use it by changing the base URL to the Nova AI Relay API endpoint and using a Nova AI Relay API key.
Which models can I use?+
You can use the models enabled in the Nova AI Relay model catalog, including GPT, Claude, Gemini, and image generation models. Model availability, context length, pricing, and capabilities may change as upstream providers update their offerings.
How is usage billed?+
Usage is billed from your Nova AI Relay credit balance based on the selected model and metered usage such as input tokens, output tokens, cached tokens, images, or requests. You can review usage, logs, and remaining credits in the user dashboard.
Where do I create an API key?+
After signing in, open the API Keys page in the Nova AI Relay dashboard. Create a key, store it securely, and use it as the bearer token for API requests. Rotate keys if they are exposed or no longer needed.
Can I use Nova AI Relay with Claude Code, Codex CLI, Gemini CLI, OpenCode, OpenClaw, or Hermes Agent?+
Yes. The documentation includes integration guides for CCSwitch, Claude Code, Codex CLI, Gemini CLI, OpenCode, OpenClaw, and Hermes Agent. Most tools work through an OpenAI-compatible or custom provider configuration.
Does Nova AI Relay train models on my prompts?+
Nova AI Relay does not use your prompts or outputs to train Nova AI Relay-owned foundation models. Requests are routed to upstream model providers, and those providers may process data according to their own terms and policies.
How do I troubleshoot failed API requests?+
Check that your API key is active, the base URL is correct, the model ID is copied exactly from the model catalog, and your account has enough credits. Then review the Logs page for status codes, latency, errors, and request details.
Does Nova AI Relay have an invitation or referral program?+
Yes. Users can share an invitation link from the dashboard. By default, when an invited user recharges, the inviter earns a long-term reward equal to 10% of that user's recharge amount, subject to the active program rules and anti-abuse checks.