Skip to content
OpenSmartRoute
All vendors
NVIDIA

Vendor

NVIDIA models

10 models in the reference catalogue, vendor list prices per million tokens. 0 are routable on this deployment today.

Models

10

0 routable here

Cheapest (blended)

free

Nemotron 3.5 Lightning (free)

Median list price

$0.04

3:1 input to output per 1M tokens

Strongest

-

no published benchmarks

Every NVIDIA model

Newest first. Prices are the vendor's published rates per million tokens; click a model for the full specification.

ModelReleasedContextInput / 1MOutput / 1MIntelligenceCapabilities
Nemotron 3.5 Lightningnvidia/nemotron-3.5-lightningAug 11, 2026262K$0.08$0.20-
reasoning tools open weights
Nemotron 3.5 Lightning (free)nvidia/nemotron-3.5-lightning:freeAug 11, 20261Mfreefree-
reasoning tools open weights
Nemotron 3.5 Content Safetynvidia/nemotron-3.5-content-safetyJun 4, 2026131K$0.20$0.20-
reasoning open weights
Nemotron 3.5 Content Safety (free)nvidia/nemotron-3.5-content-safety:freeJun 4, 2026128Kfreefree-
reasoning open weights
Nemotron 3 Ultranvidia/nemotron-3-ultra-550b-a55bJun 4, 2026262K$0.63$3.13-
reasoning tools open weights
Nemotron 3 Ultra (free)nvidia/nemotron-3-ultra-550b-a55b:freeJun 4, 20261Mfreefree-
reasoning tools open weights
Nemotron 3 Nano Omni (free)nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:freeApr 28, 2026256Kfreefree-
reasoning tools open weights
Nemotron 3 Supernvidia/nemotron-3-super-120b-a12bMar 11, 20261M$0.09$0.40-
reasoning tools open weights
Nemotron 3 Super (free)nvidia/nemotron-3-super-120b-a12b:freeMar 11, 2026262Kfreefree-
reasoning tools open weights
Nemotron 3 Nano 30B A3Bnvidia/nemotron-3-nano-30b-a3bDec 14, 2025262K$0.05$0.20-
reasoning tools open weights

Frequently asked

What is the cheapest NVIDIA model?
Nemotron 3.5 Lightning (free) at free per 1M input tokens and free per 1M output tokens (vendor list price).
Which NVIDIA model has the largest context window?
Nemotron 3.5 Lightning (free) accepts 1M tokens of context.
How do I route to NVIDIA models with OpenSmartRoute?
Declare a target in targets.yaml with metadata.model set to the catalogue id (for example nvidia/nemotron-3.5-lightning); the router scores it against every other target on cost, quality, latency and your policies for each request.

Route NVIDIA with everything else

OpenSmartRoute picks the cheapest model that meets your quality bar per request, so a NVIDIA flagship handles hard prompts while small models take the rest. Price a workload or try a routing decision live.