The Number
- DeepSeek's flash model lists at $0.15/M input cache-miss off-peak, with cache-hit input at $0.003/M, per DeepSeek's pricing docs as of 2026-10-01.
- Z.ai's glm-flash routes at $0.02/M input, the cheapest non-free China-lab listing on OpenRouter as of 2026-10-01.
- Alibaba's qwen3-max lists at 2.5 yuan/M input for the 0-32K tier, rising to 7 yuan/M at 128K-256K, per DashScope pricing as of 2026-10-01.
Top of the Fold
- Huawei ships the Mate 90 series with the Kirin 9050 Pro on top variants and claims a 31% performance gain on the Pro Max over last year, its first full-stack flagship refresh of the cycle, per 智东西.
- Huawei tops the Mate 90 Pro Max out at 18,999 yuan, pairing a detachable module camera and HarmonyOS 7 in a two-hour spec-heavy launch, per 虎嗅.
Untranslated
- Alibaba's Qwen flags an acceleration of its Personal Agent push while Tencent is said to be running three product lines and ByteDance fast-tracks its Spell project, a domestic scramble after Meta's Muse hit, per 虎嗅 (circulating, 虎嗅's read, verify at vendors).
- Qwen's lineup on OpenRouter now spans a free qwen3.8-27b tier up through qwen3.8-flash at $0.15/M, the widest single-lab price ladder among China models as of 2026-10-01 on OpenRouter.
- Otherwise quiet across the Chinese community overnight; wires and signal feeds returned little primary China-lab material inside the window.
The Kicker
- A Huawei Mate 90 flagship near the 19,000-yuan mark is already circulating in a 59-yuan building-block case, per 虎嗅.
