NVIDIA: Nemotron 3 Ultra (free)
NVIDIA Developer Architecture Profile
Intelligence (ELO)1049Chatbot Arena Verified
Max Context Limit1,000,000Max Out: 65,536 tokens
Prompt CachingStandardNo Cache Discount
Standard API / 1MFREE🧠 Deep Reasoner
Model Capabilities & Contract SLAs
- Reasoning
- Free Tier
- 🧠 Extended Test-Time Reasoning
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 (Standard Prompt)Free / 1M
Output Tokens (Completion)Free / 1M
Pricing data via OpenRouter. Sync: 10/2/2026
Evaluate Competitors
VS Engine MatchupNVIDIA: Nemotron 3 Ultra (free) vs Mistral: Voxtral Small 24B 2507VS Engine MatchupNVIDIA: Nemotron 3 Ultra (free) vs OpenAI: gpt-oss-120bVS Engine MatchupNVIDIA: Nemotron 3 Ultra (free) vs Space Bunny AlphaVS Engine MatchupNVIDIA: Nemotron 3 Ultra (free) vs Auto RouterVS Engine MatchupNVIDIA: Nemotron 3 Ultra (free) vs Meta: Muse Spark 1.3 ContributorVS Engine MatchupNVIDIA: Nemotron 3 Ultra (free) vs Tencent: Hy-MT2-7B