About China AI Hub

Last updated: 2026-10-23

What China AI Hub is

China AI Hub is an independent, English-language knowledge and intelligence platform for understanding China's AI ecosystem — models, agents, APIs, companies, technologies, pricing, benchmarks and the relationships between them. It exists to answer one practical question: which Chinese AI model, agent or API is suitable for which task, at what cost, with what capabilities, and under what deployment conditions?

Concretely, the site is four things:

A database of models, companies, agents, APIs, pricing and benchmarks;
A knowledge base of AI technologies and their Chinese implementations;
A comparison system that explains trade-offs rather than declaring winners;
A research platform publishing original analysis with clearly labeled evidence.

What China AI Hub is not

Not an official source — the site is not affiliated with, endorsed by, or operated by any AI vendor (DeepSeek, Alibaba Cloud, ByteDance, Moonshot, MiniMax, Zhipu or others).
Not a news aggregator or translation site.
Not a "Top 10 Chinese AI models" listicle.
Not a ranking site that declares one best model.
Not a collection of rewritten official documentation.
Not a provider of AI systems, APIs, or any paid service.

Why this website exists

China's AI ecosystem moves quickly and its documentation is fragmented across Chinese-language sources. International developers and researchers routinely lack a reliable, structured, English-language view of what exists, what it costs, and what evidence supports it. China AI Hub fills that gap with structured data and sourced analysis — not with hype or rankings. The full scope and limits of what the content can be relied on for are in the Disclaimer.

What is on the site

The site is organized around eleven content collections, each answering a different question:

Models — Chinese AI models and their specifications, licenses and benchmarks.
Companies — the companies behind the models, and their products.
Agents — Chinese coding and general-purpose AI agents.
API — the API platforms and their endpoints, compatibility and capabilities.
Pricing — normalized API pricing across providers.
Benchmarks — benchmark evaluations and their methodology.
Technology — a knowledge base of AI technologies with Chinese adoption context.
Comparisons — evidence-based model comparisons, without rankings.
Guides — decision frameworks for choosing and evaluating models.
Research — original analysis with labeled evidence.
News — short, sourced release notes.

Each entity page (models, companies, agents, APIs, pricing, benchmarks) is a data record with sourced fields and a verification status; the editorial pages (research, guides, comparisons, technology, news) add interpretation on top of that data. The same facts are also served in machine-readable form on the companion Data Hub (data.sinoaihub.com).

Who it serves

AI developers, software engineers, researchers, enterprise technology buyers, startups, analysts and journalists — anyone evaluating Chinese AI systems from outside China. Content assumes an intelligent reader who knows AI but may not know China's AI ecosystem.

Who operates and writes the site

The site is operated by the China AI Hub project. Research and writing are done by the SinoAI Hub Research Team — a team-level byline, not a set of named individuals. The project deliberately avoids publishing invented individual "expert" profiles; accountability comes from the source layer (every claim carries a source, a source type and a verification date), not from a persona. The project is independent and ad-free, as described in the funding disclosure. The project does not publish a corporate legal name, registration number, or registered office; the contact point is Martin.W — [email protected].

How data is sourced

Every important factual claim carries a source, a source type, a publication date and a last-verified date. Sources are ranked in a four-tier hierarchy — primary sources first (official documentation, model cards, pricing pages, papers, GitHub), then high-quality technical sources, then major media, then community — and a community-only source can never carry a critical fact alone. Where sources conflict, the conflict is explained, not silently resolved. Where data is unknown, the site says so. Nothing is invented.

The full collection hierarchy, the verification workflow, and the eight-field evidence layer are documented on the methodology page. How the site is licensed, how the main site and the Data Hub (data.sinoaihub.com) divide labor, and how often data is re-verified are on the Data Policy page.

Verification and evidence

Every entity record carries a verification status: verified or partially_verified. A record is partially_verified when a core field is documented as not publicly disclosed — the gap is stated, never hidden. Across the reference dataset this is currently a 45 / 14 split (45 records fully verified, 14 partially verified), and the evidence layer normalizes every source into eight fields — identifier, name, URL, type, publication date, verification date, confidence, and conflict flag. This is what makes the site's claims checkable rather than asserted.

Analysis policy

Editorial analysis (including the sections labeled "China AI Hub analysis") is always labeled as analysis and is traceable to cited evidence. The site does not promote or attack companies, does not declare universal winners, and does not pretend certainty where none exists. The four claim labels — Official fact, Vendor-reported claim, Third-party evidence, and China AI Hub analysis — are applied throughout so the reader can always tell whose claim they are reading. The quality gate is documented in the editorial standards.

Use of AI in producing the site

The site is transparent about where AI-assisted tooling is used. Illustrations on entity pages are AI-generated and labeled in image credits (e.g. "Seedream"). Automated tooling is used in the verification workflow to re-fetch primary sources and compare extracted values against stored values. Editorial text and analysis are produced and reviewed by the SinoAI Hub Research Team, and analysis is labeled. See the Disclaimer for the complete AI disclosure.

How to read the site

The site is built so you can tell, on any page, what kind of statement you are reading. Entity pages open with a short answer block that states what the thing is, why it matters, its key characteristics, and what a professional should know, before the detailed data. Facts carry sources and verification dates; analysis is labeled. Every record shows a verification status (verified or partially_verified), and the four claim labels — Official fact, Vendor-reported claim, Third-party evidence, and China AI Hub analysis — tell you whose claim you are reading.

Comparisons explain trade-offs without declaring a winner; decision guides route you to the best-documented fit for a scenario rather than to a "best model"; research pages state their method and their uncertainties. If a value is unknown, the page says so rather than guessing. The methodology and editorial standards pages define these conventions in full.

Editorial values

Zero fabrication — no invented prices, benchmarks, parameters, dates or company details. Unknown values are marked, not guessed.
No winner verdicts — the site explains trade-offs and suitability rather than declaring a universal best model.
Honest uncertainty — where evidence is thin or conflicting, the site says so, and does not pretend certainty.
Source over persona — credibility comes from the evidence layer, not from invented author authority.
Verification over speed — a sourced, date-stamped record beats a fast, unsourced claim.

Independence and funding

The site operates with no advertising, no sponsored listings, no paid placements and no affiliate links. No AI vendor pays for placement, ranking or description. The project is operated at its own expense and currently has no revenue model; if that changes, the funding disclosure will be updated before the change takes effect.

Corrections and accountability

Errors are handled through a public, recorded process — see the corrections policy — and site changes are logged in the update history. Data licensing and reuse terms are in the Data Policy and the Terms of Use.

How the main site and the Data Hub relate

The project deliberately splits content across two properties so that analysis and raw data stay separable. The main site (sinoaihub.com) is the editorial authority: it explains what things mean, why they matter, and how to choose between them. The Data Hub (data.sinoaihub.com) is the reference layer: it carries the exact, machine-readable facts and their evidence records. Every Data Hub record links back to its canonical main-site page, and the two layers are kept in agreement by automated cross-checks. Interpretation lives on the main site; the Data Hub is not a second blog.

How the site stays current

Data changes on three cadences, and staleness is shown rather than hidden: light weekly re-checks of pricing and status for top models and providers; a monthly full crawl of all monitored targets; and event-driven updates within 24 hours of a major release, price cut or license change. Every record shows a last_verified date — under 30 days counts as current, 30–90 days is flagged "may be outdated", and over 90 days is marked prominently for verification before reliance. The full schedule is in the Data Policy, and the chronological record of changes is in the update history.

Contact

To reach the project: Martin.W — [email protected], or see the contact page for all channels.

The site's operating rules and disclosures live alongside this page: Methodology (how data is collected and verified), Editorial Standards (the quality gate), Data Policy (licensing and update cadence), Authors (who writes the site), Funding Disclosure (how the project is funded), Corrections (how errors are handled), Updates (the change history), Terms of Use, Privacy Policy, Cookie Policy, and Disclaimer.