Compare AI models

Specifications
Parameters
340B110B
Context window
4k32k
Benchmarks
HumanEval
73.2%52.4%
MMLU
78.7%80.4%
GSM8K
92.3%85.4%
Overview
CompanyNVIDIAQwen
Release dateJun 14 2024Apr 25 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

Nemotron-4 340B leads Qwen1.5-110B on 2 of the 3 benchmarks they both report (HumanEval, MMLU, GSM8K). Qwen1.5-110B shipped 50 days before Nemotron-4 340B, so benchmark comparisons should account for the intervening progress.

Nemotron-4 340B has 340B parameters, while Qwen1.5-110B has 110B. Context windows are 4k (Nemotron-4 340B) vs 32k (Qwen1.5-110B).

On HumanEval, Nemotron-4 340B leads at 73.2% vs Qwen1.5-110B at 52.4%. On MMLU, Qwen1.5-110B leads at 80.4% vs Nemotron-4 340B at 78.7%. On GSM8K, Nemotron-4 340B leads at 92.3% vs Qwen1.5-110B at 85.4%.