NVIDIA: Nemotron 3 Ultra (batch)

NVIDIA Developer Architecture Profile

Intelligence (ELO)1437Chatbot Arena Verified

Max Context512,288Tokens

API Cost / 1M$2.10Blended Prompt + Completion

Model Capabilities

Coding & Logic
Fictional

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...

Granular Pricing Matrix

Input Tokens (Prompt)$0.30 / 1M

Output Tokens (Completion)$1.80 / 1M

Pricing data via OpenRouter. Sync: 8/6/2026

Evaluate Competitors

VS Engine MatchupNVIDIA: Nemotron 3 Ultra (batch) vs Z.ai: GLM 5.2 (batch)VS Engine MatchupNVIDIA: Nemotron 3 Ultra (batch) vs Qwen: Qwen3.6 27B VS Engine MatchupNVIDIA: Nemotron 3 Ultra (batch) vs DeepSeek: DeepSeek V4 Flash 0423 VS Engine MatchupNVIDIA: Nemotron 3 Ultra (batch) vs Qwen: Qwen3.5-27B VS Engine MatchupNVIDIA: Nemotron 3 Ultra (batch) vs Mistral Large 2407 VS Engine MatchupNVIDIA: Nemotron 3 Ultra (batch) vs WizardLM-2 8x22B