The Information Machine

SemiAnalysis tested Qwen3.8-Max-Preview 2.4T params https://t.co/iN84UKxB4b

SemiAnalysis Twitter · SemiAnalysis (@SemiAnalysis_) · 2026-08-02

SemiAnalysis announces it has tested Qwen3.8-Max-Preview, a 2.4 trillion parameter mixture-of-experts model from Alibaba's Qwen team, with results available at a linked URL.

Open original ↗

Appears in

Kimi K3 and Qwen3.8-Max: Chinese Labs Close Gap with Closed Frontier Models

Extraction

Topics: qwenllm-evaluationlarge-language-modelsmixture-of-experts

Claims

SemiAnalysis conducted performance testing of Qwen3.8-Max-Preview, a 2.4 trillion parameter model.
Qwen3.8-Max-Preview represents one of the largest publicly tested language models by parameter count.

Key quotes

SemiAnalysis tested Qwen3.8-Max-Preview 2.4T params