Qwen 3 Max (Preview) — benchmark results

Alibaba's preview release of Qwen 3 Max. Provider: Alibaba. Released 2025-09-05. Access: API.

Unified ELO 1663 ± 13, rank #336 of 1776 rated models, from 53 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Wolfram LLM Benchmarking Project61.5Correct Functionality (%)91
AA MMLU-Pro83.83Accuracy (%)87.5
Chatbot Arena (Text)1435Elo80.1
Vals AI TaxEval v273.96Accuracy (%)73.8
AA Omniscience - Humanities & Social Sciences27.4Accuracy (%)73.4
AA Omniscience - Software Engineering (SWE) - Java25Accuracy (%)73
AA Omniscience - Law17.9Accuracy (%)72.9
AA Omniscience - Software Engineering (SWE) - PHP38Accuracy (%)72.8
AA Omniscience - Software Engineering (SWE) - Python31.5Accuracy (%)71.9
AA LiveCodeBench65.08Pass@1 (%)71.8
AA Omniscience - Software Engineering (SWE) - C50Accuracy (%)70.8
AA AIME 202575Accuracy (%)70.6

Interactive version: theaggregate.ai/model?slug=qwen-3-max-preview · How the rankings work · Data refreshed daily, snapshot 2026-07-22.