Qwen2 72B Instruct vs Qwen3-235B-A22B

Alibaba / Qwen

Qwen2-72B-Instruct is a 72-billion-parameter language model released by Alibaba's Qwen team in June 2024, serving as the flagship of the Qwen2 generation and representing a major step in open-weight multilingual modeling. Trained on data spanning 30+ languages with strong coverage of code and structured reasoning, the model was among the first openly released 70B-class models to demonstrate competitive performance across diverse benchmarks. It established the foundation architecture and training methodology that the Qwen2.5 series would later extend.

Alibaba / Qwen

Qwen3-235B-A22B, released by Alibaba's Qwen team on April 28, 2025, is a Mixture-of-Experts large language model with 235 billion total parameters and 22 billion active parameters per inference. It features a 256K token context window, hybrid thinking capabilities (both reasoning and direct generation modes), and was trained on 36 trillion tokens across 119 languages. Qwen3-235B targets complex reasoning, multilingual tasks, and open-source deployments under the Apache 2.0 license.