nvidia/nemotron-3-ultra-550b-a55b:cyberrouter
NVIDIA Nemotron 3 Ultra is an open frontier-reasoning and orchestration model from NVIDIA, with 55B active parameters out of 550B total (MoE). Built on a hybrid Transformer-Mamba mixture-of-experts architecture, it... — security reference context with source traces and usage accounting.
- Context window
- 1,000,000 tokens
- Input / 1M tokens
- free
- Output / 1M tokens
- free
Published catalog snapshot in USD; prices and availability can change. Live endpoints and your account access are checked separately.
Loading endpoints…
Prices are per million tokens, converted from the per-token strings the gateway returns. Amber marks a number that is an estimate rather than a charge.
Reading endpoints, providers and keys…
Context length, full pricing, capabilities and the measured constraints for each tag are in the endpoint list above. platform: not servable means the platform cannot route that endpoint today — the endpoint or its provider is not marked live, or we hold no credential for it. It is never a statement about your key: an endpoint we cannot serve is reachable the moment you bring one of your own.
Joined in this browser from /v1/models/nvidia/nemotron-3-ultra-550b-a55b/endpoints, /v1/providers and /v1/keys/byok. Only the last needs your key, and it returns masked rows — the console never holds a provider secret.