Compare AI models

Specifications
Parameters
16B120B
Context window
32k1M
API pricing
Cheapest input
—$0.06DeepInfra
Cheapest output
—$0.32DeepInfra
—$35.00DeepInfra · bf16
These models have no shared benchmark scores.
Overview
CompanyDeepSeekNVIDIA
Release dateMay 16 2024Mar 11 2026
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek V2 Lite and Nemotron 3 Super don't publish scores on any of the same benchmarks, so there's no direct head-to-head comparison. DeepSeek V2 Lite shipped 664 days before Nemotron 3 Super, so benchmark comparisons should account for the intervening progress.

DeepSeek V2 Lite has 16B parameters, while Nemotron 3 Super has 120B. Context windows are 32k (DeepSeek V2 Lite) vs 1M (Nemotron 3 Super).

Direct benchmark comparisons are unavailable — DeepSeek V2 Lite and Nemotron 3 Super don't publish scores on any of the same benchmarks.