Compare AI models

Specifications
Parameters
21B>1T
Context window
128k256k
API pricing
Cheapest input
$0.018AkashML—
Cheapest output
$0.09AkashML—
$8.10Darkbloom · fp8—
Benchmarks
SWE-Bench Verified
60.7%75.3%
Humanity's Last Exam · with tools
17.3%49.8%
GPQA Diamond
71.5%87.4%
Overview
CompanyOpenAIQwen
Release dateAug 5 2025Jan 26 2026
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

Qwen3-Max-Thinking leads gpt-oss-20b on 3 of the 3 benchmarks they both report (SWE-Bench Verified, Humanity's Last Exam, GPQA Diamond). gpt-oss-20b shipped 174 days before Qwen3-Max-Thinking, so benchmark comparisons should account for the intervening progress.

gpt-oss-20b has 21B parameters, while Qwen3-Max-Thinking has >1T. Context windows are 128k (gpt-oss-20b) vs 256k (Qwen3-Max-Thinking). gpt-oss-20b is open weight, while Qwen3-Max-Thinking is closed.

On SWE-Bench Verified, Qwen3-Max-Thinking leads at 75.3% vs gpt-oss-20b at 60.7%. On Humanity's Last Exam · with tools, Qwen3-Max-Thinking leads at 49.8% vs gpt-oss-20b at 17.3%. On GPQA Diamond, Qwen3-Max-Thinking leads at 87.4% vs gpt-oss-20b at 71.5%.