Qwen3 235B

Alibaba's flagship 235B MoE model

Qwen3 235B is Alibaba's most capable model, using a mixture-of-experts architecture with 235 billion total parameters and 22 billion active. It delivers top-tier performance on multilingual reasoning, coding, and math tasks.

Try Qwen3 235B

Specifications

Context window128k tokens

Input price$0.130 / 1M tokens

Output price$0.600 / 1M tokens

Credits per query5 cr

Released2025-04

LicenseQwen License

Features

Tool useJSON modeFunction callingCode

Benchmark Scores

What do these mean?

Sourced from official model cards and academic papers. Higher is better.

GPQA Diamond

PhD-level science & reasoning

—

AIME 2024

Mathematical olympiad problems

85.7%

Arena Elo

Chatbot Arena human preference ranking

1303

HumanEval

Python coding — pass@1 rate

91%

MATH

Competition-level mathematics

96.7%

MMLU

General knowledge across 57 academic domains

89%

MT-Bench

Multi-turn instruction following (out of 10)

—

SWE-Bench

Real-world GitHub issue resolution rate

—

More from Qwen

Qwen3 32B3 cr

Dense 32B model with hybrid thinking

Qwen3 VL 235B5 cr

Frontier open-source vision-language model

Qwen2.5 VL 72B3 cr

Leading open-source vision-language model