Compare AI models

Specifications
Parameters
8B/70B/405B1.5B/7B/72B
Context window
128k4k
API pricing
Cheapest input
$0.02DeepInfra—
Cheapest output
$0.04DeepInfra—
$6.00DeepInfra · fp8—
Benchmarks
MATH
73.8%85.9%
GSM8K
96.8%95.9%
Overview
CompanyMetaQwen
Release dateJul 23 2024Sep 19 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

LLaMA 3.1 and Qwen2.5-Math are evenly matched across the 2 benchmarks they both report (MATH, GSM8K). LLaMA 3.1 shipped 58 days before Qwen2.5-Math, so benchmark comparisons should account for the intervening progress.

LLaMA 3.1 has 8B/70B/405B parameters, while Qwen2.5-Math has 1.5B/7B/72B. Context windows are 128k (LLaMA 3.1) vs 4k (Qwen2.5-Math).

On MATH, Qwen2.5-Math leads at 85.9% vs LLaMA 3.1 at 73.8%. On GSM8K, LLaMA 3.1 leads at 96.8% vs Qwen2.5-Math at 95.9%.