Alibaba's biggest model ever is going open-weights Alibaba (@AlibabaGroup) has unveiled Qwen3.8-Max, its most capable model yet at 2.4T parameters, and for the first time a Max-class Qwen goes open-weights, with the download landing next week. Priced at $2 per million input

What happened

Alibaba unveiled Qwen3.8‑Max, its most capable model with two point four trillion parameters, and announced it will be released as open‑weights software that can be downloaded next week, priced at two dollars per million input tokens.

Confirmed

Global impact / market context

Open‑weights models let any developer access the technology without a licensing fee, which can speed up product development, lower costs for firms needing large language models, and broaden AI adoption across industries.

Analyst inference

The launch adds a new low‑cost competitor to the generative‑AI market dominated by firms charging higher fees, potentially pressuring pricing and encouraging cloud providers to offer more affordable AI compute options.

Analyst inference

What to watch

1. The speed at which startups and large enterprises begin using Qwen3.8‑Max, showing how quickly the model captures market share and influences AI spending patterns. Proposed
2. Reactions from rival AI providers, such as adjustments to their pricing or feature offerings, if Alibaba’s two‑dollar per million token rate attracts significant user interest. Proposed
3. Any regulatory developments concerning large AI models, especially rules on data privacy or model safety, that could affect how quickly Alibaba can roll out Qwen3.8‑Max. Proposed

Evidence

• x.com