Compare AI models

Specifications
Parameters
—14.3B
Context window
8k32k
API pricing
Cheapest input
$30.00Azure—
Cheapest output
$60.00Azure—
$9,000Azure—
Benchmarks
HumanEval
67%34.2%
MMLU
86.4%62.5%
GSM8K
92%61.5%
Overview
CompanyOpenAIQwen
Release dateMar 14 2023Mar 28 2024
AccessClosedOpen Weight
Model detailsView modelView model

Frequently asked questions

GPT-4 leads Qwen1.5-MoE on 3 of the 3 benchmarks they both report (HumanEval, MMLU, GSM8K). GPT-4 shipped 380 days before Qwen1.5-MoE, so benchmark comparisons should account for the intervening progress.

Context windows are 8k (GPT-4) vs 32k (Qwen1.5-MoE). GPT-4 is closed, while Qwen1.5-MoE is open weight.

On HumanEval, GPT-4 leads at 67% vs Qwen1.5-MoE at 34.2%. On MMLU, GPT-4 leads at 86.4% vs Qwen1.5-MoE at 62.5%. On GSM8K, GPT-4 leads at 92% vs Qwen1.5-MoE at 61.5%.