--- title: "Why Are US Startups Routing 30%+ of Tokens to Chinese AI Models? The Inside-China Read" date: 2026-08-26 originalDate: 2026-08-08 originalTitle: "谁在给硅谷换AI底座？8周5款中国大模型正在改写规则" issue: "Issue 2" description: "Five Chinese flagship models in eight weeks — the inside-China read on why US startups route 30%+ of tokens to them. Translated, sourced, [unverified] tagged." tags: - "Model releases" - "OpenRouter" - "Qwen" - "DeepSeek" - "GLM" - "Kimi" sources: - label: "The Register: China turns up the heat with open model blitz (2026-08-03)" url: "https://www.theregister.com/ai-and-ml/2026/08/03/china-turns-up-the-heat-with-open-model-blitz-as-us-model-makers-panic/5282526" - label: "Reuters: Alibaba unveils its most capable AI model (2026-08-03)" url: "https://www.reuters.com/business/retail-consumer/alibaba-unveils-its-most-capable-ai-model-date-not-far-behind-moonshots-size-2026-08-03/" - label: "China Daily: Zhejiang AI models gain ground — DeepSeek V4 Pro GA (2026-08-19)" url: "https://regional.chinadaily.com.cn/ezhejiang/2026-08/19/c_1206425.htm" faq: - q: "Which Chinese AI models launched in the summer 2026 wave?" a: "Per our translated essay: GLM-5.2 (mid-June, topped CodeArena), Kimi K3 (July 16, 2.8T parameters), DeepSeek-V4-Flash (late July), Qwen3.8-Max (early August, 2.4T-parameter MoE), and ByteDance's Seedance 2.5 (early August). Reuters and The Register independently cover the same August wave." - q: "What share of usage do Chinese models have on OpenRouter?" a: "Per the essay citing OpenRouter data [unverified]: in July 2026 the five most-called models on OpenRouter were all Chinese, and US enterprises' token consumption on Chinese models had risen from 4.5% eighteen months earlier to a stable 30%+, peaking at 46%." - q: "Why are US companies switching to Chinese AI models?" a: "The translated take: cost and openness. The essay cites ~$25 (Claude) vs ~$0.18 (DeepSeek-V4) for the same coding workload, and names Lindy (API costs down ~90%), Coinbase (AI spend nearly halved; 91% of engineers don't need frontier performance), and Tenstorrent (5× cost cut). The pattern it calls 'layered mixing': Chinese models for routine 80% of work, US flagships for the hard 20%." --- In the eight weeks ending August 2026, five Chinese flagship models launched in succession — and for the first time, the essay claims, the five most-called models on OpenRouter were all Chinese, despite 47% of the platform's users being US developers [unverified]. US enterprises' token consumption on Chinese models reportedly rose from 4.5% eighteen months ago to a stable 30%+, peaking at 46% [unverified], as companies like Coinbase and DoorDash shifted routine workloads to cheaper, open-weight Chinese models. ## The numbers - 8 weeks, 5 flagship models - OpenRouter: top 5 most-called models all Chinese (July 2026) [unverified] - 47% of OpenRouter users are US developers; 6% Chinese - US-enterprise token share on Chinese models: 4.5% → 30%+ stable since February 2026, peak 46% [unverified] - GLM-5.2: mid-June, topped CodeArena; Kimi K3: July 16, 2.8T parameters; DeepSeek-V4-Flash: late July; Qwen3.8-Max: early August, 2.4T-parameter MoE; Seedance 2.5: early August - Stanford 2026 AI Index: US–China top-model performance gap narrowed to 2.7% [unverified] - Lindy (25-person startup): API costs down ~90% after migrating from Claude to DeepSeek-V4; bills previously exceeded total payroll - Coinbase: AI spending nearly halved; internal survey — 91% of engineers don't need frontier-level performance - Tenstorrent: costs down 5× after switching from Claude - Cost comparison: ~$25 (Claude) vs ~$0.18 (DeepSeek) for the same coding workload; some Chinese models cost ~1% of US closed flagships - Qwen APP update August 7: free research features, scheduled tasks, PC-controlling office assistant ## The take inside China The story Chinese tech media are telling: a release blitz with no precedent. Eight weeks, five flagships — covering text, code, multimodal, and video — while OpenAI and Anthropic update flagships on a yearly cadence. The essay's argument: this isn't brute-force compute. It's a different technical bet — MoE architecture plus inference-side optimization plus open weights — instead of one giant closed model recouped via high API prices. The payoff shows in adoption: the named switchers are telling (Lindy, Coinbase, DoorDash, Tenstorrent), and the candid counterpoint gets airtime too: US engineers note gaps remain in extreme deep reasoning. The real pattern, the essay argues, is "layered mixing" — Chinese models for the 80% of routine work, US flagships for the hard 20%. Its pointed question: if China eats 80% of workload volume, can the remaining 20% sustain US closed-model valuations? Meanwhile, the Qwen APP update on August 7 — free research, scheduled tasks, a PC-controlling office assistant — takes the model layer straight to ordinary users; viral Excel-and-PPT demos on Douyin and Bilibili did marketing no benchmark could. ## Why it matters outside If the cost gap (reportedly up to two orders of magnitude) and the open-weight advantage hold, Western AI pricing and business models face real pressure — and enterprises gain leverage via multi-vendor hedging. The trajectory also complicates the assumption that export controls can hold back Chinese frontier AI. English coverage of the same wave: [The Register](https://www.theregister.com/ai-and-ml/2026/08/03/china-turns-up-the-heat-with-open-model-blitz-as-us-model-makers-panic/5282526), [Reuters](https://www.reuters.com/business/retail-consumer/alibaba-unveils-its-most-capable-ai-model-date-not-far-behind-moonshots-size-2026-08-03/), [China Daily](https://regional.chinadaily.com.cn/ezhejiang/2026-08/19/c_1206425.htm). ## Sources - [The Register: China turns up the heat with open model blitz](https://www.theregister.com/ai-and-ml/2026/08/03/china-turns-up-the-heat-with-open-model-blitz-as-us-model-makers-panic/5282526) - [Reuters: Alibaba unveils its most capable AI model](https://www.reuters.com/business/retail-consumer/alibaba-unveils-its-most-capable-ai-model-date-not-far-behind-moonshots-size-2026-08-03/) - [China Daily: Zhejiang AI models gain ground](https://regional.chinadaily.com.cn/ezhejiang/2026-08/19/c_1206425.htm) > **Provenance & disclosure.** Originally published in Chinese on our WeChat channel on 2026-08-08 ("谁在给硅谷换AI底座？8周5款中国大模型正在改写规则"); drafted with AI assistance under human editorial direction. Translated to English on 2026-08-26 (AI-assisted, human-reviewed). The OpenRouter share figures, named-company quotes, and the Stanford 2.7% figure are the essay's citations, kept with [unverified] tags; the release wave itself is corroborated by the English sources above. The essay carries a pro-China framing; that framing is preserved as translated opinion, not asserted as fact. This is translated commentary — not a SigPulse measurement. Our first-party measurements live in the [dispatches](/posts/) and the [/data/ ledger](/data/).