Nemotron 3 Ultra
by NVIDIAtext->text4 endpoints
Neura Intelligence Index
59.4/ 100
Rank
#115—
Confidence
mediumBenchmarks
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
Modalities
Input
Text
Output
Text
Technical Specifications
Context Window
1M tokens
Tokenizer
Other
Uptime (24h)
9960.1%
Parameters
550000000000B
License
openmdwOpen Weights
Country
Supported Parameters
frequency_penaltyinclude_reasoninglogit_biasmax_tokensmin_ppresence_penaltyreasoningreasoning_effortrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_p
Benchmark Performance
coding
| Benchmark | Score | Source |
|---|---|---|
| SWE-bench Verified | 70.7 | verified |
| SciCode | 44.6 | verified |
reasoning
| Benchmark | Score | Source |
|---|---|---|
| GPQA Diamond | 87 | verified |
| Humanity's Last Exam |
Performance Over Time
Pricing
Per 1M tokens
Input$0.60
Output$3.60
Cache read$0.20