Pilot publication · Working title · No public deployment yet
English-language China AI intelligence

From original source to practical consequence.

A weekly professional briefing that connects Chinese primary sources, technical change, market structure, policy context, and the best specialist analysis elsewhere.

Latest issue

Kimi K3 lands
as WAIC opens.

What is available now, what is still promised, and what builders should watch before treating a product launch as a deployable open-weight release.

Issue 0July 17, 20269 min read

“Released” is not one state.

Kimi K3 is a live product and API. Its full model weights are promised next. The distinction changes what application teams, self-hosters, infrastructure providers, and policy readers should do today.

API LIVE WEIGHTS PENDING HOLD SELF-HOSTING JUDGMENT
The brief

Four signals.
Four different actions.

Not every important development needs a long article. Each card separates the fact, the consequence, and the next useful action.

01 / BUILD

Kimi K3: product live, weights later

Evaluate the API now. Defer self-hosting conclusions until artifacts, license, and runtime support exist.

02 / BUILD

DeepSeek model-ID deadline

Legacy aliases are scheduled to end July 24. Hard-coded IDs and fallback assumptions need inspection.

03 / BUILD

Qwen Code moves up the stack

Automatic model fallback and nested sub-agents make workflow reliability—not only model quality—the competitive unit.

04 / POLICY

WAIC links industry and diplomacy

Open and accessible AI is becoming part of China's international technology offer. Adoption will be the harder evidence.

Signature method

Source. Context.
Consequence.

The publication does not compete by claiming exclusive access to every topic. It competes by making a fragmented market easier to understand and use.

S

Source

Start with the primary document, product page, technical note, filing, transcript, or policy text. Separate what is directly established from what is reported or inferred.

C

Context

Connect the event to the preceding model, company, policy, and market history. Compare how the story is framed inside and outside China.

C

Consequence

Translate the event into the question a professional reader actually faces: what to build, test, buy, monitor, migrate, or reassess.

Inside China

What the company, developer community, policy text, and domestic market emphasize.

Outside China

How global media, investors, builders, and policymakers frame the same event.

What actually matters

The operational synthesis after the two narratives are placed against the primary evidence.

Built for work

One briefing.
Several professional lenses.

The audience is not “everyone interested in AI.” It is people whose work becomes worse when China AI is understood late, incompletely, or through a single narrative.

Builders

Model, product, and infrastructure teams deciding what to test, integrate, or migrate.

Markets

Founders, investors, and strategists tracking companies, products, capital, and competitive structure.

Policy

Researchers and decision-makers who need technical and commercial facts behind policy narratives.

China-facing teams

Global professionals who cannot continuously monitor Chinese-language primary sources.

Weekly pilot

Start with the source.
End with the consequence.

One concise English briefing for professionals who need to understand China AI without monitoring Chinese sources full time.

Prototype form. No address is collected in this offline build.