This entry is a serving variant of NVIDIA: Nemotron 3 Ultra. A variant shares its base model's tokenizer by construction, so the measurement lives on the base record and this page inherits it.
| Signal | Result | Reference |
|---|---|---|
| Tokenizer signature | Not measured | – |
| API surface | No other entry shares this exact contract | api_d0c2ec50 |
| Context and output | 512,288 context · – max output | declared |
| Reasoning contract | Optional · efforts high, medium · default high | declared |
| Serving providers observed | None recorded yet | – |
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...
Supported parameters
frequency_penaltyinclude_reasoninglogit_biasmax_tokensmin_ppresence_penaltyreasoningreasoning_effortrepetition_penaltyresponse_formatstopstructured_outputstemperaturetool_choicetoolstop_ktop_p
defaults: {"temperature":1,"top_p":0.95,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null}
Cite this page as the evidence record for NVIDIA: Nemotron 3 Ultra (batch): the URL is stable, measurements are dated, and revisions append to the history above. Method: tokenizer fingerprinting, v1, 50 strings. The raw data behind every figure is on the data page.
