Compare AI models

Specifications
Parameters
671B51B
Context window
128k128k
API pricing
Cheapest input
$0.259SiliconFlow—
Cheapest output
$0.38DeepInfra—
$71.00DeepInfra · fp4—
These models have no shared benchmark scores.
Overview
CompanyDeepSeekNVIDIA
Release dateDec 1 2025Sep 23 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

DeepSeek-V3.2 and Llama 3.1 Nemotron 51B don't publish scores on any of the same benchmarks, so there's no direct head-to-head comparison. Llama 3.1 Nemotron 51B shipped 434 days before DeepSeek-V3.2, so benchmark comparisons should account for the intervening progress.

DeepSeek-V3.2 has 671B parameters, while Llama 3.1 Nemotron 51B has 51B. Context windows are 128k (DeepSeek-V3.2) vs 128k (Llama 3.1 Nemotron 51B).

Direct benchmark comparisons are unavailable — DeepSeek-V3.2 and Llama 3.1 Nemotron 51B don't publish scores on any of the same benchmarks.