Login
NVIDIAPremiumPremiumActive

NVIDIA: Nemotron 3 Ultra

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...

Text
Start with this modelBrowse all models
Specification
Model ID
nvidia/nemotron-3-ultra-550b-a55b
Modality
Text >text
Context
262.1K tokens
Input
$0.600/1M
Output
$2.40/1M
Updated
Sep 20, 2026
Scores
Capabilities
  • Supports tool calling
  • No vision support listed
  • Supports long-context workflows
  • Supports deeper reasoning tasks

Input and output

Accepts Text inputs. Returns Text outputs.

Supported parameters

frequency_penaltyinclude_reasoninglogit_biasmax_tokensmin_ppresence_penaltyreasoningreasoning_effortrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_p

Related models

DeepSeek

DeepSeek: DeepSeek Pro Latest

This model always redirects to the latest model in the DeepSeek Pro family.

Inference.net

Inference.net: Schematron V2 Small

Schematron V2 Small is a 3B-parameter HTML-to-JSON extraction model from Inference.net. It prioritizes extraction quality for complex schemas and long pages. Extraction instructions must be supplied through a JSON schema...

Inference.net

Inference.net: Schematron V2 Turbo

Schematron V2 Turbo is a 3B-parameter HTML-to-JSON extraction model from Inference.net. It prioritizes throughput for high-volume extraction workloads. Extraction instructions must be supplied through a JSON schema in response_format rather...