🔥 Qwen 4 teases frontier power on a diet
Alibaba just previewed Qwen 3.8-Flash-Next, a day early, and the specs say it hits near-frontier scale on a fraction of the compute. This isn't a routine model drop, it's a signal that the efficiency war just got real. China is optimizing for cost-per-intelligence while the US still burns cash on scale. Cheaper frontier-adjacent models mean margin compression for every closed-model API business.
Watch who ships Qwen 4 first, and watch OpenAI's pricing page the same week.
1August 26, 2026 137 1
