Zhipu AI official site: how this independent reference relates to the upstream lab

What this site is — and what it is not

A direct statement on the nature and purpose of zai.gr.com, and the distinction between this independent reference and the upstream Zhipu AI official site.

zai.gr.com is an independent reference site. It summarises publicly available information about Z.ai (Zhipu AI), the GLM model family, the BigModel developer platform, and the surrounding open-source ecosystem. Nothing on this site is produced by Zhipu AI, reviewed by Zhipu AI, or published with Zhipu AI's knowledge or consent. The relationship is analogous to an independent product wiki or a third-party documentation mirror — useful as a reader-first summary, not authoritative as a company statement.

The zhipu ai official site — the surface operated directly by Zhipu AI — is a distinct property. It is where the company publishes its own product announcements, official documentation, terms of service, and pricing pages. Content there reflects Zhipu AI's own editorial decisions and should be treated as authoritative for contractual, legal, and pricing questions. Content here reflects this site's editorial decisions about how to present publicly available information in a way that is useful to an English-speaking developer audience.

The two surfaces serve different purposes and different reader needs. The upstream official site is the right destination for signing up, reading the current terms of service, checking the latest pricing, and accessing official support. This independent reference is the right destination for understanding the model landscape at a glance, comparing GLM tiers, finding integration patterns, and navigating the relationship between the brand names (Z.ai, Zhipu AI, BigModel, ChatGLM, THUDM) that can otherwise be confusing for outside teams.

The upstream Z.ai and Zhipu AI surfaces

Four upstream surfaces make up the official Zhipu AI presence. Each has a distinct purpose and audience.

The first is the Z.ai consumer surface — the primary public-facing website and chat application that operates under the Z.ai brand. This is where most end users interact with the GLM models through a browser or mobile app without any API configuration. It is the right starting point for anyone who wants to use Z.ai conversationally without developer setup.

The second is the BigModel developer console. This is the official developer surface for API key management, billing, model selection, and usage dashboards. Developers who need programmatic access to the GLM models sign up here, generate API keys here, and track their spend here. It is the authoritative source for current pricing numbers, API endpoint documentation, and rate-limit specifications.

The third is the THUDM GitHub organisation. This is the official open-source home for ChatGLM inference code, GLM fine-tuning toolkits, and evaluation harnesses. When the zhipu ai official site references open-source releases, the canonical code lives in THUDM repositories.

The fourth is the THUDM Hugging Face organisation. This is where the open-weight model files are published. For the downloadable ChatGLM and GLM builds, Hugging Face under the THUDM namespace is the authoritative source. Community quantisations and mirrors appear under other namespaces and are third-party derivatives, not official releases.

Zhipu AI source surfaces — what each one is and what to expect from it

Source type	Purpose	What to expect
This site (zai.gr.com)	Independent English-language reference and developer orientation	Reader-first summaries of public information; not authoritative for contracts, pricing, or support
Upstream Z.ai consumer surface	Official Zhipu AI product site and chat application	Company announcements, terms of service, official pricing, account creation, direct chat access
BigModel developer console	Official developer platform for API keys, billing, and model management	Authoritative pricing figures, API documentation, rate limits, and account management
Hugging Face / THUDM mirror	Official open-weight model hosting under THUDM namespace	Authoritative weight files, model cards, and license terms for downloadable GLM builds

How to verify you are on the right surface

A three-step check distinguishes the upstream official surfaces from independent resources like this one.

First, check the domain. The upstream Z.ai consumer surface and the BigModel console operate under specific domains that Zhipu AI has announced in official contexts — on the THUDM GitHub README files, in company blog posts, and in the API documentation. zai.gr.com is not one of those domains. The .gr.com suffix indicates this is a third-party reference site, not a first-party Zhipu AI property.

Second, check the footer of any page you are reading. Every page on this site carries the disclaimer "Independent reference resource on Z.ai (Zhipu AI) — Not affiliated with Zhipu AI / BigModel / ChatGLM project." Official Zhipu AI properties do not carry this disclaimer. If a footer says this is independent, it is independent.

Third, for any contractual, legal, or support question, go directly to the upstream official surface. This site does not have access to Zhipu AI's internal systems, cannot answer support queries on Zhipu AI's behalf, and cannot make commitments about pricing or availability. For those needs, the upstream BigModel developer console is the right destination. Guidance from ai.gov on AI vendor evaluation and due diligence is a useful frame for teams formalising their procurement process for any AI API provider.

Why an independent reference is useful

Independent references serve a distinct function that official product sites are not designed for — they organise information around reader intent rather than product marketing.

Official product sites are excellent at onboarding a user who has already decided to adopt the product. They walk through sign-up, show the product's own framing of its strengths, and document the API contract in detail. What they are less optimised for is helping a developer who is still in evaluation mode — who needs to understand the relationship between brand names, compare model tiers without marketing framing, find integration patterns for a specific framework, or understand the download and self-hosting path without starting an account.

This site is built for that evaluation mode. The content is organised by reader question rather than by product structure. The tone is neutral — this site does not advocate for or against any particular model or platform, including Z.ai. The goal is a reference that a developer can read in 15 minutes and come away with an accurate map of the Z.ai ecosystem, after which they can make their own decision about whether to engage with the official surfaces.