Back to Models
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...
6/4/2026
512,288 tokens
#77 Text (german)
Specifications
Modalities
Input
text
Output
text
Supported Parameters
frequency_penalty
include_reasoning
logit_bias
max_tokens
min_p
presence_penalty
reasoning
reasoning_effort
repetition_penalty
response_format
seed
stop
structured_outputs
temperature
tool_choice
tools
top_k
top_p
Max Output Tokens
16,384Reasoning Configuration
- Default
- Thinking on (effort: high)
- Selectable effort levels
- highmedium
- Turning thinking off
- Possible (reasoning.enabled: false)
Data policy
- Prompt retention
- Unknown — we could not confirm
- Training
- Unknown — not declared
"Unknown" does not mean "safe". It means we could not confirm it.
Whether a provider trains on prompts is a declared value from our terms with them. Retention is determined from the upstream listing for every host this model can reach. Neither is guessed.
Frequently asked questions
- Is Nemotron 3 Ultra available on FastMetal?
- Not at the moment. Nemotron 3 Super, from the same lab, is available on the FastMetal API today.
- What is the context window of Nemotron 3 Ultra?
- 512,288 tokens, shared between the prompt and the response.
- How does Nemotron 3 Ultra rank?
- #122 on the public arena's Overall board (ELO 1,426). Ranks move as the leaderboard is updated.
Leaderboard
Text
russianELO: 1,402
#150CodingELO: 1,473
#121MathELO: 1,444
#88Creative WritingELO: 1,382
#143Instruction FollowingELO: 1,405
#140Hard PromptsELO: 1,444
#128Multi-TurnELO: 1,398
#169OverallELO: 1,426
#122ChineseELO: 1,464
#118KoreanELO: 1,351
#138EnglishELO: 1,440
#116frenchELO: 1,448
#114germanELO: 1,448
#77spanishELO: 1,424
#116Rankings: LMArena leaderboard dataset (lmarena-ai/leaderboard-dataset), CC BY 4.0, snapshot 2026-10-22; category names mapped and ratings rounded by FastMetal.