nvidia Open weights Reasoning

Nemotron 3 Ultra pricing & benchmarks

Released Jun 4, 2026 · served by 4 providers · ranked #36 by intelligence in our index

Input / 1M tokens$0.500
Output / 1M tokens$2.20
Context window512K
Cached input$0.100
Intelligence index37.8
Coding index49.3
Agentic index27.4
1M in + 300K out$1.16

What Nemotron 3 Ultra is

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...

Providers and prices for Nemotron 3 Ultra

The same model costs different amounts depending on who serves it. Prices are per 1M tokens, cheapest first. Uptime and throughput are OpenRouter's rolling measurements, not vendor claims.

ProviderInputOutputContext QuantUptime 24hThroughput
DeepInfracheapest $0.500 $2.20 262K fp4 94.9%
BaseTen $0.600 $2.40 203K fp4 98.7%
Venice $0.625 $3.13 256K fp8 91.0%
Together $0.600 $3.60 512K 96.6%

Nemotron 3 Ultra benchmark results

Design Arena head-to-head results, as reported through the OpenRouter model API.

CategoryArenaRankEloWin rate
svg models #49 1128 37.6%
3d models #50 1190 41.8%
asciiart models #50 1112 37.5%
uicomponent models #59 1180 39.5%
gamedev models #60 1183 38.8%
codecategories models #69 1155 36.2%
dataviz models #71 1151 37.4%
website models #82 1127 32.5%

Cheaper models in the same class

Models scoring within 4 points of Nemotron 3 Ultra on the Intelligence Index, but with a lower output price.

FAQ

How much does the Nemotron 3 Ultra API cost?

$0.500 per 1M input tokens and $2.20 per 1M output tokens. A typical workload of 1M input + 300K output tokens costs about $1.16.

Is Nemotron 3 Ultra free?

No. It is a paid model starting at $0.500 per 1M input tokens, though some providers offer trial credits.

Which provider is cheapest for Nemotron 3 Ultra?

DeepInfra at $0.500 input / $2.20 output per 1M tokens (fp4 quantization).

Can I self-host Nemotron 3 Ultra?

Yes — weights are published as nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16 on Hugging Face, so you can run it on your own hardware.