EastVoice · AI · 2026-09-21 · Issue 74

Editor's Note: China's open-source wave keeps crashing outward. StepFun unveiled Step 5 Preview, a 600B-parameter sparse MoE that already sits in the global top three among open models, with full weights promised on October 15. Alibaba open-sourced Qwen-Image-2.1, and a Beijing ministry official called open-source foundation models national industrial policy.

Commentary: The money story is stranger than the model story. Unitree shed ¥250B / ~$35B in market cap within a month of listing, while the primary market threw ¥93.5B / ~$13B at embodied AI in the same half-year. Secondary markets now punish what venture capital happily funds.

---

I. AI Foundation Models

1. Zhipu's MaaS platform lets enterprises demand zero data retention

• Summary: Zhipu AI said its model-serving platform is rolling out a no-content-retention mode for enterprise and developer customers. Once activated, inputs and outputs are not persisted server-side; any user can apply for it.
• Source: ITHome · 2026-09-20
• Editor's Take: Privacy guarantees are becoming a billing feature in China's model wars. When pricing converges, the compliance sheet is the new leaderboard.

2. Alibaba open-sources Qwen-Image-2.1, a unified generate-and-edit image model

• Summary: Alibaba's Qwen team released Qwen-Image-2.1, merging text-to-image and image editing in one model, supporting transparent-image output and up to 10 reference images per generation. The team pitches it as the best quality-per-cost model in its open image series.
• Source: ITHome · 2026-09-20
• Editor's Take: Reference-image conditioning at this scale is how Chinese labs close the gap with closed rivals, one weight dump at a time. Editing and generation in one backbone is what Western labs still ship as two products.

3. StepFun's Step 5 Preview lands in the global open-model top three

• Summary: StepFun announced flagship Step 5 Preview, a sparse MoE with 600B total parameters aimed at coding, software engineering, professional work and finance. It already ranks top-three among open models on Artificial Analysis, and StepFun committed to releasing full weights on October 15.
• Source: ITHome · 2026-09-20
• Editor's Take: A preview that beats released open models is a marketing move, but the October 15 date is the real commitment. Watch whether they actually ship on that day — Chinese labs have been burned on open-source promises before.

4. China Telecom puts full-stack domestic AI office software on a single RTX 3090

• Summary: China Telecom demoed an enterprise AI office suite built entirely on domestic models that runs on one consumer RTX 3090, letting companies keep documents and prompts inside their own network. It enters a summer arms race where big Chinese firms all chase the AI-office entry point.
• Source: QbitAI · 2026-09-20
• Editor's Take: The pitch isn't intelligence, it's air gap. Sovereignty sells in China the way low latency sells elsewhere, and a telco bundling models with private lines is structurally hard to undercut.

5. APUS open-sources one of the first cross-platform reproductions of Jev

• Summary: Shanghai-based APUS said its AI lab published among the earliest independent open-source reproductions of the newly released Jev model, demonstrating sub-second decision latency on domestic models.
• Source: QbitAI · 2026-09-20
• Editor's Take: Reproducing a hot model within days is the open-source ecosystem's real speed test. China's labs are no longer just consumers of frontier releases; the turnaround clock now runs in their favor.

6. MIIT vice minister: open-source base and vertical models are national industrial policy

• Summary: At the World Manufacturing Convention in Hefei, MIIT vice minister Xin Guobin said the ministry will push open-source foundation models and vertical-domain models as core tools, with "AI + manufacturing" as the main attack direction for the next phase of industrial policy.
• Source: ITHome · 2026-09-20
• Editor's Take: When a ministry names open source in a keynote, procurement follows. Expect state-backed enterprises to start writing "must run on open weights" into tenders.

II. AI Software & Applications

7. Tencent's WorkBuddy takes both consumer and enterprise desktop-agent crowns, says Frost & Sullivan

• Summary: Frost & Sullivan's 2026 desktop AI agent market report, released September 20, ranks Tencent's WorkBuddy first in China on both the personal and enterprise desktop-agent charts.
• Source: Leiphone · 2026-09-20
• Editor's Take: WeChat, Docs and meeting data all feed one assistant — distribution beats benchmark again. The number one spot in agents in China is really a distribution scoreboard.

8. Tencent upgrades Yuanbao's AI recorder with simultaneous record-and-shoot capture

• Summary: Yuanbao app V2.85.0 adds record-while-capturing, internal device audio, custom editing, and one-tap sharing into WorkBuddy and Tencent Docs.
• Source: ITHome · 2026-09-20
• Editor's Take: The recorder is a trojan horse. Every meeting transcript auto-filed into Tencent Docs feeds the same corpus WorkBuddy retrieves from — that's the strategy in one feature note.

9. Alibaba's Qwen app adds a teen mode that unlocks only study features

• Summary: Qwen upgraded its minor-protection mode: teen accounts keep photo homework help, AI writing, translation and deep research, while responses are age-filtered, chatbot boundaries are spelled out, and usage-time reminders are enforced.
• Source: ITHome · 2026-09-20
• Editor's Take: Regulator-facing, almost certainly. But a whitelisted study-only mode is also a wedge: the kid with the Qwen homework habit is the adult with the Qwen subscription.

10. SiliconFlow closes Series B+ and C rounds, near ¥2.9B / ~$400M raised in 2026 alone

• Summary: The domestic model-serving cloud completed its second B+ tranche and a Series C, bringing equity funding this year to roughly ¥2.9 billion / ~$400 million.
• Source: ITHome · 2026-09-20
• Editor's Take: Inference serving is where China's open-weight glut monetizes. More open models means more customers who need someone else to run them — SiliconFlow sells the shovel after the gold rush gave away the gold.

11. ByteDance's spun-off drug-discovery arm lands $290M first round at ~$1.5B valuation

• Summary: Anew Labs, spun out of ByteDance's AI drug line in June, closed a $290 million first institutional round at roughly $1.5 billion post — the largest single AI-pharma round in China this year. Its fastest pipeline asset is still preclinical.
• Source: TMTPost · 2026-09-20
• Editor's Take: A preclinical company at unicorn pricing tells you investors are valuing the compute and the data engine, not the molecule. That's either visionary or very early Dot-com behavior, and history says you can't tell yet.

12. 107 startup deals this week as robot components slip into global supply chains

• Summary: Weekly China venture tallies recorded 107 financing events, with a clear tilt toward humanoid-robot parts makers — actuators, dexterous hands, sensors — landing orders inside global OEM supply chains.
• Source: Huxiu · 2026-09-20
• Editor's Take: The boring layer keeps winning. Nobody shorted a component supplier — the boring parts trade beats the glamorous humanoid demo on cash flow.

13. Unitree sheds ¥250B / ~$35B since IPO while primary market still pays any price

• Summary: Unitree Robotics hit ¥444.9B market cap on its first STAR Market day, August 19, then lost ¥250 billion within a month. Embodied-AI startups meanwhile raised about 322 rounds totaling ¥93.5B / ~$13B in H1 2026, a fivefold year-over-year increase.
• Source: Huxiu · 2026-09-20
• Editor's Take: This is the cleanest dataset China's AI boom has produced: public markets rate commercialization, private markets rate narrative, and the gap between the two is the industry's real valuation.

14. Embodied AI's star executives keep resigning at the peak of the funding wave

• Summary: Over six months of record raises — Unitree's listing, Agibot's HK IPO push, Qianxun's single ¥1B round, Xinghaitu's ¥2B B+ — a parallel outflow has gone underreported: founding technologists and commercial leads leaving their own companies, per a GlowStream survey of the sector.
• Source: Huxiu · 2026-09-20
• Editor's Take: Money arrives, founders exit. The most honest bearish signal in a bubble is who declines to spend the new round at your company.

15. The data-collection center boom starts to show cracks

• Summary: Dozens of locally backed embodied-AI "data collection centers" proliferated across China; founder criticism from Mech-Mind's Shao Tianlan of deal-stacking has drawn attention, and reporting notes many centers now struggle to prove their training data gets used.
• Source: Huxiu · 2026-09-20
• Editor's Take: Subsidized teleoperation farms producing data nobody trains on is a very specific failure mode of industrial policy. The robots were real; the dataset economics were fiction.

16. AI app founders confront the growth math problem: more users, faster losses

• Summary: At a closed Hard AI 100 session covered by Huxiu, Chinese AI-application founders described shrinking feature moats — a capability built this quarter gets commoditized by the next model release — and unit economics where usage growth accelerates losses.
• Source: Huxiu · 2026-09-20
• Editor's Take: This is the applied layer's original sin everywhere, and Chinese founders are saying it out loud because their model supply is open-source and free, which strips away the last excuse: there is no moat but distribution and vertical data.

III. Humanoid Robots

17. "World's first personal robot": Agibot's Qiyuan Q1 goes on sale from ¥19,999 / ~$2,800

• Summary: At its global product launch on September 20, Qiyuan Robotics (backed by "Bumblebee He") priced the Q1 humanoid at ¥19,999, with an Explorer edition at ¥26,999 — consumer-electronics pricing for a full humanoid.
• Source: ITHome · 2026-09-20
• Editor's Take: Sub-$3,000 is the price point where a humanoid becomes a developer console instead of a museum piece. Whatever Q1 does at launch, the hackable install base is the asset.

18. Qiyuan's pitch: sell the personal robot into ordinary life first

• Summary: Ifanr's hands-on with the ¥19,999 Qiyuan T1 argues the company is doing the opposite of the industry: instead of promising future domestic robots, it is shipping companion, telepresence and camera-tracking functions people will actually pay for today.
• Source: ifanr · 2026-09-20
• Editor's Take: Every embodied-AI deck promises the kitchen; Qiyuan is selling the living room. Beachhead companionship into later utility is at least a plan, which is more than most demo-videos companies have.

19. Faraday Future launches nine EAI robot configurations, topping out above ¥920,000 / ~$130,000

• Summary: Faraday Future held a September 19 launch event in the US unveiling its "quad-core full-intelligence" EAI robot line, the All-New Futurist, Master Mini and FX Aegis series — nine configurations, the most expensive priced above ¥920,000.
• Source: ITHome · 2026-09-20
• Editor's Take: An EV company that never shipped cars at scale is now shipping the priciest humanoid SKU on the board. The market should price the track record, not the configurator.

20. Dongfeng Motor sends humanoid robots into its plants in October

• Summary: Dongfeng's chief intelligent-technology engineer Zhang Zhenlin said humanoid robots will begin sorting and quality-inspection work at Dongfeng factories this October, with small-batch trials by year-end and a stated goal of human-parity capability by end of 2027. The company's robot-dog program is also in development.
• Source: ITHome · 2026-09-20
• Editor's Take: A state automaker deploying its own robots, on its own timeline, for its own QA line is the least glamorous and most credible humanoid story in China. Self-hosting beats selling.

21. Levink Robotics bets robots should teach themselves

• Summary: GeekPark profiled Levink's self-evolution approach, in which robots improve manipulation from their own deployed operation data rather than centrally curated teleoperation datasets.
• Source: GeekPark · 2026-09-20
• Editor's Take: Self-evolving fleets vs. data farms is the embodied-AI training-strategy fork of the year. Whichever compounds on deployment data wins the cost curve — the data centers' business model only works if the robots don't.

22. JD pushes its stir-fry robot from delivery kitchens into home kitchens at ¥19,888 / ~$2,800

• Summary: JD.com launched "Qixian Dachu", a family healthy-meal service built around a cooking robot, opening pre-orders in Beijing, Shanghai, Guangzhou and Shenzhen with a ¥19,888 founding-member price.
• Source: Huxiu · 2026-09-20
• Editor's Take: The robot is the marquee; the recurring meal-kit subscription is the business. JD is buying its way into the kitchen through the back door of a hardware one-time charge.

IV. Autonomous Driving

23. Stelato RX adds an 81 kWh pack priced from ¥269,800 / ~$38,000, with an L3 architecture trim

• Summary: Harmony Intelligent Mobility Alliance opened pre-orders for a new battery variant of the Stelato RX crossover: lidar-equipped cabin trim from ¥269,800, and a trim with L3 autonomous-driving architecture from ¥329,800.
• Source: ITHome · 2026-09-20
• Editor's Take: Huawei's pricing sheet now lists an L3 badge as a trim level. Whether that's regulatory readiness or marketing, buyers will treat it as the former.

24. BAIC's new boxy 5X confirms Huawei Qiankun smart-driving fitment

• Summary: Beijing Auto Works unveiled studio photos of the Xingtan 5X off-road-styled SUV and officially confirmed it carries Huawei's Qiankun driver-assistance stack.
• Source: ITHome · 2026-09-20
• Editor's Take: Huawei's answer to Android is licensing the driving stack to everyone who can't build one. The more badges, the stronger the moat — as long as the cloud keeps paying.

25. "Phantom braking" is becoming Chinese smart-driving's biggest complaint

• Summary: A widely shared Guokr piece catalogs owners' frustration with ADAS that slams the brakes on open roads with no visible cause — "the hardest part of smart driving was never making the car go, it's teaching it to stop" — as phantom braking complaints cluster across Chinese forums.
• Source: Huxiu · 2026-09-20
• Editor's Take: Conservative braking is statistically safe and experientially enraging, and trust is the actual product being sold here. Recall rates won't capture churn of this.

V. World Models / Physical AI

26. Coscene's Boundless world model posts two global-first metrics on WorldArena 2.0

• Summary: On the final WorldArena 2.0 leaderboard, Coscene's Boundless world model took second globally on Track 1 video quality while ranking first on background consistency and JEPA similarity — its claim to long-horizon scene coherence leadership.
• Source: Leiphone · 2026-09-20
• Editor's Take: Consistency metrics are where world models get honest — visual fidelity flatters, coherence over minutes betrays. A Chengdu company leading those is worth more attention than the demo reels suggest.

---

Frontiers of the Physical World | EastVoice