Qwen 3.8 Max vs Gemini 3.5 Pro: AI Model Showdown
TL;DR
Alibaba Labs released Qwen 3.8 Max, adding a new dynamic to its competition with proprietary models such as Google DeepMind's Gemini 3.5 Pro. The model is cited at 2.4 trillion parameters with notable cost efficiency, processing input tokens at $2 per million. That pushes price per token, not just raw benchmark performance, to the center of the comparison. For high-volume teams it is a number worth running through the math.
Nauti's Take
The real breakthrough in this news is $2 per million input tokens, because at volume price decides more than the last benchmark point. The open question is how stable Qwen 3.8 Max stays on long contexts and non English text, where Gemini has been dependable.
Teams running millions of tokens should test Qwen as a second model before moving production load.
Video
Sources