Compare AI models

Specifications
Parameters
8B/70B/405B14.3B
Context window
128k32k
API pricing
Cheapest input
$0.02DeepInfra—
Cheapest output
$0.04DeepInfra—
$6.00DeepInfra · fp8—
Benchmarks
HumanEval
89%34.2%
MMLU
87.3%62.5%
GSM8K
96.8%61.5%
Overview
CompanyMetaQwen
Release dateJul 23 2024Mar 28 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

LLaMA 3.1 leads Qwen1.5-MoE on 3 of the 3 benchmarks they both report (HumanEval, MMLU, GSM8K). Qwen1.5-MoE shipped 117 days before LLaMA 3.1, so benchmark comparisons should account for the intervening progress.

LLaMA 3.1 has 8B/70B/405B parameters, while Qwen1.5-MoE has 14.3B. Context windows are 128k (LLaMA 3.1) vs 32k (Qwen1.5-MoE).

On HumanEval, LLaMA 3.1 leads at 89% vs Qwen1.5-MoE at 34.2%. On MMLU, LLaMA 3.1 leads at 87.3% vs Qwen1.5-MoE at 62.5%. On GSM8K, LLaMA 3.1 leads at 96.8% vs Qwen1.5-MoE at 61.5%.