FrontierAugust 5, 2026via AI News
Alibaba, DeepSeek push China’s AI model race towards lower costs
Why it matters
Two Chinese labs are redefining the frontier benchmark: model size and capability per unit of cost. This matters to practitioners choosing between vendors and to the lab-race narrative — the cost-per-inference efficiency angle shifts where the competitive pressure lands.
Key signals
- Alibaba Qwen3.8-Max: 2.4 trillion parameters, mixture-of-experts architecture
- 95 billion parameters active per request (MoE efficiency)
- DeepSeek V4-Flash inference pricing lower than competing systems
- Positioning around cost-efficiency as competitive lever, not just raw capability
The hook
Alibaba's Qwen3.8-Max (2.4T parameters, MoE) and DeepSeek's V4-Flash are forcing the frontier race toward efficiency — not just capability.
Alibaba has launched Qwen3.8-Max, its largest AI model to date, as DeepSeek’s latest V4-Flash model draws attention for inference pricing that is lower than several competing systems. Qwen3.8-Max has 2.4 trillion parameters and uses a mixture-of-experts architecture, which activates only part of the…