Compare AI models

Specifications
Parameters
1B/4B/12B/27B14.3B
Context window
—32k
API pricing
Cheapest input
$0.04DeepInfra—
Cheapest output
$0.05Parasail—
$10.50Parasail · bf16—
Benchmarks
HumanEval
87.8%34.2%
MMLU
76.9%62.5%
GSM8K
95.9%61.5%
Overview
CompanyGoogleQwen
Release dateMar 12 2025Mar 28 2024
AccessOpen WeightOpen Weight
Model detailsView modelView model

Frequently asked questions

Gemma 3 leads Qwen1.5-MoE on 3 of the 3 benchmarks they both report (HumanEval, MMLU, GSM8K). Qwen1.5-MoE shipped 349 days before Gemma 3, so benchmark comparisons should account for the intervening progress.

Gemma 3 has 1B/4B/12B/27B parameters, while Qwen1.5-MoE has 14.3B.

On HumanEval, Gemma 3 leads at 87.8% vs Qwen1.5-MoE at 34.2%. On MMLU, Gemma 3 leads at 76.9% vs Qwen1.5-MoE at 62.5%. On GSM8K, Gemma 3 leads at 95.9% vs Qwen1.5-MoE at 61.5%.