Compare AI models

Specifications
Parameters
—0.5B/1.5B/7B/57B/72B
Context window
100k128k
Benchmarks
HumanEval
71.2%86%
MMLU
78.5%82.3%
GSM8K
88%91.1%
Overview
CompanyAnthropicQwen
Release dateJul 11 2023Jun 6 2024
AccessClosedOpen Weight
Model detailsView modelView model

Frequently asked questions

Qwen2 leads Claude 2 on 3 of the 3 benchmarks they both report (HumanEval, MMLU, GSM8K). Claude 2 shipped 331 days before Qwen2, so benchmark comparisons should account for the intervening progress.

Context windows are 100k (Claude 2) vs 128k (Qwen2). Claude 2 is closed, while Qwen2 is open weight.

On HumanEval, Qwen2 leads at 86% vs Claude 2 at 71.2%. On MMLU, Qwen2 leads at 82.3% vs Claude 2 at 78.5%. On GSM8K, Qwen2 leads at 91.1% vs Claude 2 at 88%.