SF·SIX / MODEL PROFILE
NVIDIA 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...
NVIDIA Nemotron 3 Ultra costs $0.50 per 1M input tokens and $2.20 per 1M output tokens at the standard tier, the lowest default-region price among the 3 providers SF·SIX tracks, from DeepInfra (fp4). Cached input reads cost $0.10 per 1M tokens. Updated ; prices checked .
- Developer
- NVIDIA
- Input modalities
- text
- Output modalities
- text
- Context tokens
- 262,144
- Latest price observation
- 2026-10-17
- Lowest paired standard token quote
- $0.50 input / $2.20 output per 1M tokens
Dated price and provider records
| Provider and channel | Input | Output or native rate | Unit and status | Source and observation |
|---|---|---|---|---|
| DeepInfraOpenRouter | $0.50 | $2.20 | 1M tokenspriced · USD | OpenRouter2026-10-17 |
| BasetenOpenRouter | $0.60 | $2.40 | 1M tokenspriced · USD | OpenRouter2026-10-17 |
| Venice AIOpenRouter | $0.625 | $3.13 | 1M tokenspriced · USD | OpenRouter2026-10-17 |
| BasetenOpenRouter · US region | $0.60 | $2.40 | 1M tokenspriced · USD | OpenRouter2026-10-17 |
| DeepInfraOpenRouter · Flex tier | $0.40 | $1.76 | 1M tokenspriced · USD | OpenRouter2026-10-17 |
Batch and free variants
- Nemotron 3 Ultra (free) · from $0.00 input / $0.00 output per 1M tokens
These are dated, non-executable observations. Verify the exact endpoint, price and terms with the source before use. Unknown values are not free.