Compare AI models

Specifications
Parameters
—0.6B/1.7B/4B/8B/14B/30B/32B/235B
Context window
128k—
API pricing
Cheapest input
—$0.064DeepInfra
Cheapest output
—$0.192DeepInfra
—$30.00DeepInfra · fp8
Benchmarks
LiveCodeBench
62.5%70.7%
MATH-500
96.2%98%
AIME 2024
77.5%85.7%
Overview
CompanyMoonshot AIQwen
Release dateJan 20 2025Apr 29 2025
AccessClosedOpen Weight
Model detailsView modelView model

Frequently asked questions

Qwen3 leads Kimi K1.5 on 3 of the 3 benchmarks they both report (LiveCodeBench, MATH-500, AIME 2024). Kimi K1.5 shipped 99 days before Qwen3, so benchmark comparisons should account for the intervening progress.

Kimi K1.5 is closed, while Qwen3 is open weight.

On LiveCodeBench, Qwen3 leads at 70.7% vs Kimi K1.5 at 62.5%. On MATH-500, Qwen3 leads at 98% vs Kimi K1.5 at 96.2%. On AIME 2024, Qwen3 leads at 85.7% vs Kimi K1.5 at 77.5%.