Provider failover
Requests retry across healthy upstream providers and keys. If one channel fails, the next one takes over automatically — no code changes on your side.
One OpenAI-compatible endpoint for 18 frontier models. Provider failover, streaming responses and per-token billing — so you ship AI features, not infrastructure.
curl https://YOUR-DOMAIN/v1/chat/completions \
-H "Authorization: Bearer $NOVA_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "glm-5.3-flash",
"messages": [{ "role": "user", "content": "Hello, Nova!" }]
}'
Trusted by engineering teams at
We handle the provider mess — capacity, outages, rate limits — behind one stable endpoint, so your product never waits on a model vendor.
Requests retry across healthy upstream providers and keys. If one channel fails, the next one takes over automatically — no code changes on your side.
Requests automatically reroute to the next-best healthy provider mid-outage. Zero code changes, zero dropped requests.
Per-token pricing with no minimums and no seats. Volume discounts kick in automatically as you scale.
Drop-in compatible with the OpenAI SDK. Change one base URL and keep every line of your existing code.
Live throughput and per-token pricing measured across our network. Pick a model manually, or let auto-routing choose the fastest healthy provider.
| Model | Context | Output speed | Media | Input / 1M | Output / 1M | Actions |
|---|
Prices are per 1M tokens on standard tier. Latency measured from us-east-1 over the last 24h.
Grab an API key, send your first request and stream tokens from 18 models. Founding free plan for the first 50 customers — no credit card required.