Story perspectives

Alibaba Unveils 2.4T AI Model, Ranks Second in Image Analysis

8/3/2026

Summary

Story summary

Alibaba unveiled the 2.4-trillion-parameter AI model Qwen3.8-Max on Aug 3, 2026.
Qwen3.8-Max processes text, images and video, handles up to 1 million tokens, and placed second worldwide for image analysis behind a Claude Fable 5 variant.
Alibaba reported benchmark scores that match or exceed Anthropic’s Fable 5 and completed a 16-day autonomous software-engineering test.
Alibaba will open-source the weights the week of Aug 10, 2026.
Alibaba’s Hong Kong shares rose up to 7.3 percent on Aug 3.

1 of 1

Story summary

Alibaba unveiled the 2.4-trillion-parameter AI model Qwen3.8-Max on Aug 3, 2026.
Qwen3.8-Max processes text, images and video, handles up to 1 million tokens, and placed second worldwide for image analysis behind a Claude Fable 5 variant.
Alibaba reported benchmark scores that match or exceed Anthropic’s Fable 5 and completed a 16-day autonomous software-engineering test.
Alibaba will open-source the weights the week of Aug 10, 2026.
Alibaba’s Hong Kong shares rose up to 7.3 percent on Aug 3.