Hub/Models/nvidia/nemotron-3-ultra-550b-a55b
nvidiatext → text

NVIDIA: Nemotron 3 Ultra

Canonical ID: nvidia/nemotron-3-ultra-550b-a55b

Compare ModelOpen on OpenRouter →

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...

Artificial Analysis Benchmarks

Verified
Intelligence Index11.8 / 100
Engineering Index22.8 / 100
Strategy & Ops15.3 / 100
Finance & Accounting24.2 / 100
Legal Index22.6 / 100
Healthcare & Medical18 / 100

Token Pricing & Economics

Prompt Input (1M Tokens)
$0.50
Input
Completion Output (1M Tokens)
$2.20
Output
Prompt KV-Cache Read (1M Tokens)
$0.10
Cached
Context Window Capacity
256k tokens
Context

Supported Parameters & Features

frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_ppresence_penaltyreasoningreasoning_effortrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p