NVIDIA

Nemotron 3 Ultra

by NVIDIAtext->text4 endpoints
Neura Intelligence Index
60.1/ 100
Rank
#132
Confidence
medium
Benchmarks
5(13% coverage)

Overview

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...

Capabilities

Text generation
Tool use / Function calling
Structured output
Extended reasoning
Prompt caching
Long context
Top-P sampling
Stop sequences
Deterministic seed

Modalities

Input
Text
Output
Text

Technical Specifications

Context Window
262.1K tokens
Max Output
32.8K tokens
Tokenizer
Other
Uptime (24h)
10000.0%
Parameters
550000000000B
License
openmdwOpen Weights
Country
🇺🇸US
Release Date
June 4, 20263mo

Supported Parameters

frequency_penaltyinclude_reasoninglogit_biasmax_tokensmin_ppresence_penaltyreasoningreasoning_effortrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_p

Benchmark Performance

coding

BenchmarkScoreSource
SWE-bench Verified
70.7
verified
SciCode
44.6
verified

reasoning

BenchmarkScoreSource
GPQA Diamond
87
verified
Humanity's Last Exam
37.4
verified

agent

BenchmarkScoreSource
BrowseComp
44.4
verified

Performance Over Time

Pricing

Per 1M tokens
Input$0.63
Output$3.13
Cache read$0.19
Added
June 4, 2026
Last synced 9/18/2026