models
Artem Vysotsky
Qwen3.8 Max Is on Writingmate: Testing Alibaba's 2.4-Trillion-Parameter Flagship on Agentic Coding and Tool Use
Alibaba's Qwen3.8 Max went GA on August 3 with 2.4 trillion parameters, a 1M-token context window, and a claim of parity with the frontier. I ran it through the same broken-function, messy-refactor, and multi-step agentic test suite used on DeepSeek V4 Pro, Mistral Large 3, and Muse Spark 1.1.
Read Article


