What is GLM and how it can help you be more productive

Z.AI assistant guide · Updated September 2026

GLM isn’t just “China’s other open-weight model” anymore. What began as bilingual research inside Tsinghua University in 2021 has grown into Z.AI (formerly Zhipu AI): a frontier open-weight lab, the world’s first publicly listed large-language-model company, and a model family that now runs inside third-party coding agents and thousands of self-hosted deployments. On August 14, 2026 Z.AI shipped GLM-5.3 – a post-training-only upgrade that lifted coding scores by half and added cybersecurity skills the company compares to Anthropic’s Mythos line.

In this guide we explain what GLM is in 2026, how the GLM-5.3 and GLM-5.3-Flash models work, what ZCode and the GLM Coding Plan can do, what the plans cost, and how to use GLM well.

Mar 2023First public release, ChatGLM chatbot
GLM-5.3 · FlashCurrent flagship · 743B MoE · Aug 14, 2026
ZCode · Coding PlanAgentic coding + subscription access
01 · What it is

What is GLM?

GLM is the family of large language models built by Z.AI, a Beijing company founded in June 2019 by Tsinghua University professors Tang Jie and Li Juanzi as a commercial spin-out of Tsinghua’s Knowledge Engineering Group, led by CEO Zhang Peng. GLM (General Language Model) research began at Tsinghua in 2021 around a bilingual Chinese-English pretraining architecture, at a time when most Western labs treated Chinese as an afterthought. The company traded as Zhipu AI domestically for years before rebranding internationally as Z.AI in July 2025, and it became the world’s first publicly listed large-language-model company on the Hong Kong Stock Exchange in January 2026.

GLM-130B (August 2022) proved the bilingual approach at scale; ChatGLM (2023) turned it into a public chatbot; and since mid-2025 Z.AI has shipped the GLM-4 and GLM-5 generations on a roughly quarterly cadence, alongside CogView/CogVideoX image and video models and the AutoGLM phone-and-computer-use agent line. Unlike OpenAI or Anthropic, Z.AI has a long habit of releasing model weights to the public – which made GLM one of the most-downloaded open model families on Hugging Face.

Today, depending on the plan and surface, GLM can:

  • Answer questions & explain concepts
  • Write, rewrite, summarize & translate
  • Search the live web with citations
  • Reason step by step with effort levels
  • Process 1M-token codebases & docs
  • Write, debug, test & ship code as an agent
  • Audit code & reason about vulnerabilities
  • Analyze images, video & documents (Flash)
  • Operate a phone or browser (AutoGLM)
  • Run locally on your own hardware via open weights
GLM – the models

GLM’s current generation is GLM-5.3, released August 14, 2026 on the exact same base model as GLM-5.2: a ~743-billion-parameter MoE (~40B active), 1M-token window, text-only, with always-on reasoning. Beside it sits GLM-5.3-Flash (August 26) – a newly trained 320B/18B base with native multimodal input under the MIT license.

GLM – the products

You meet GLM through the Z.AI chat app and API, the BigModel developer platform, the GLM Coding Plan subscription, ZCode (desktop coding agent), AutoGLM for phone and computer automation, and downloadable weights on Hugging Face, ModelScope, and GitHub.

02 · The model family

GLM-5.3, GLM-5.3-Flash, and the GLM-5 line

GLM-5.3 is unusual among frontier releases: Z.AI reused the exact same base model as GLM-5.2 and put the entire improvement into scaled post-training – more coding environments, longer tasks, stronger verifiers, more reinforcement-learning compute. In the company’s own words: “scaling post-training is all we did for GLM-5.3.” That makes it a controlled experiment in how far a frontier base can be pushed without another pretraining run.

Flagship · 743B MoE · Aug 14, 2026 (weights Aug 28)

GLM-5.3

Same base architecture as GLM-5.2 (~743B total / 40B active across 256 experts), 1M-token context, up to 128K output, text-only. Three thinking-effort levels – low, high, max (default; thinking can no longer be switched off). Z.AI-reported 50% coding lift and state-of-the-art open results on Terminal-Bench 3.0 and CyberGym.

Best for long-horizon coding & securityFlagship
Multimodal · 320B/18B active · Aug 26, 2026

GLM-5.3-Flash

The natively multimodal sibling – text, image, video, and file input – that spent a week topping usage charts under the alias “Ox Alpha” before Z.AI claimed it. Same 1M-token window, hybrid attention borrowing Kimi Delta Attention, standard unconditional MIT license.

Fast, cheap, vision-capable · fully open (MIT)
Long-horizon base · Jun 13, 2026

GLM-5.2

The 1M-context foundation GLM-5.3 builds on, with strong open-weight coding and agent results. Still widely deployed and self-hosted; on the Coding Plan, requests for it now auto-route to GLM-5.3.

Proven base · MIT weights
Earlier 2026 · Feb–Apr 2026

GLM-5 / GLM-5.1

GLM-5 (Feb 2026) scaled the line to ~744B/40B active with DeepSeek-style sparse attention; GLM-5.1 (Apr 2026) concentrated on sustained autonomous execution. Both remain available on the API.

Prior flagships · budget options

Two notes on the wider lineup. First, GLM-5.3’s weights arrived August 28 under a bespoke GLM-5.3 license – MIT-style terms for almost everyone, except that model-as-a-service operators above $10B in trailing revenue must pass a Z.AI security review first. Second, behind GLM-5 sit GLM-4.7, GLM-4.6, and GLM-4.5/GLM-4.5-Air (GLM-4.7-Flash and GLM-4.5-Air are free or near-free on the API), plus the separate GLM-4.5V/GLM-4.6V vision-language line and the CogView/CogVideoX image and video generators.

03 · How it works

How does GLM work?

GLM models are Mixture-of-Experts transformers: GLM-5.3 activates only about 40 billion of its 743 billion parameters per token, which is how a model that size can be served this cheaply. Starting with GLM-5, Z.AI adopted a DeepSeek-style sparse attention mechanism to keep the 1M-token window affordable, and GLM-5.3-Flash goes a step further by borrowing Kimi Delta Attention from Moonshot AI. Reasoning is controlled by an effort setting rather than a separate model line: GLM-5.3 always thinks, at low, high, or max effort.

Cyber capability was the surprise. Z.AI reported that as post-training scaled, security skills grew faster than expected – moving from vulnerability identification to constructing complete exploitation chains – which is why Z.AI delayed open weights by about two weeks after launch for a safety review, unlike GLM-5.2’s same-day MIT weights. A typical request flows through several stages:

  • 1

    Understand the request

    Reads the prompt, conversation history, uploaded files, and any connected tools or MCP servers. With 1M context you can feed an entire codebase.

  • 2

    Reason through the problem

    The effort setting (low / high / max) controls how much the model deliberates before answering – GLM-5.3 can no longer skip thinking entirely.

  • 3

    Decide which tools to use

    Web search, code execution, file analysis, or handing off to AutoGLM for on-device phone or browser actions.

  • 4

    Generate the response or take the action

    The answer, the pull request, the vulnerability report, or the completed multi-app task.

  • 5

    Refine with follow-ups

    Long agentic sessions inside ZCode or third-party coding agents stay coherent across the full 1M-token window.

04 · What you can do

What can you do with GLM?

Research and information

Web search grounds answers in current results with citations instead of only training data.

Writing and translating

Drafting, rewriting, summarizing, and translation – with particularly strong bilingual Chinese-English work, reflecting GLM’s original design goal.

Coding and DevOps

Runs as an agent inside ZCode and third-party coding tools; Z.AI-reported state-of-the-art among open-weight models on Terminal-Bench 3.0 and SWE-Marathon. Built for project-level work: planning across a repo, writing, testing, and iterating until CI passes.

Long-document analysis

A 1M-token context window holds an entire codebase or a book-length document in a single prompt – no aggressive summarization required.

Data analysis

Upload spreadsheets, CSVs, and PDFs; GLM cleans data, computes metrics, and explains results.

Images, video, and slides

CogView and CogVideoX generate images and video; the Flash model reads images, video, and files natively; a slide agent turns research into structured decks.

Self-hosting

MIT-licensed weights (GLM-5.3-Flash and most of the GLM-4 line) or GLM-5.3’s custom license run on your own infrastructure, keeping data off any third party’s servers entirely.

Phone and computer use

AutoGLM, open-sourced in December 2025, operates real apps – ordering food, booking travel, drafting documents – by reading the screen and simulating taps and swipes.

05 · From chatbot to agent

The agentic shift: ZCode, the Coding Plan, and open weights

Z.AI’s biggest shift mirrors rivals: from “you ask, it answers” to “you define a goal, it ships code.” ZCode, launched July 2, 2026 as an agentic development environment, is the official harness for GLM-5.3 – and unusually, the same models also power third-party agents, so the ecosystem compounds instead of fragmenting.

01

ZCode

Free to download for macOS, Windows, and Linux. An agentic IDE that reads repositories, edits files, runs tests, and manages Git for long-horizon tasks – tuned to GLM-5.3’s tool loop, with 20+ agent tools built in.

Free download · revenue via Coding Plan

02

GLM Coding Plan

Points quota with off-peak calls at a reduced coefficient and promo multipliers for subscribers – Lite, Pro, and Max tiers for steady daily coding-agent use. Calls to legacy GLM-5.2, GLM-5.1, and GLM-4.7 auto-route to GLM-5.3 or Flash.

Lite · Pro · Max · off-peak discounts

03

API & open weights

OpenAI-compatible API plus downloadable checkpoints on Hugging Face. Self-hosting removes per-token cost and data-residency concerns for regulated or high-volume workloads – the strategy is fewer surfaces that each do more, backed by weights anyone can download.

OpenAI-compatible · Hugging Face · ModelScope

The strategy is one flagship model scaled via post-training, one cheap multimodal Flash for visual coding, and one IDE that keeps everything in the same loop – with open weights so enterprises are not locked into a single vendor’s cloud.

06 · Key moments

GLM timeline: 2019–2026

Jun 2019

Tang Jie and Li Juanzi found Zhipu AI in Beijing as a commercial spin-out of Tsinghua University’s Knowledge Engineering Group.

2021–2022

GLM research begins at Tsinghua KEG; GLM-130B (August 2022), a 130-billion-parameter bilingual model, is released open-source – the architecture’s first proof at scale.

2023

ChatGLM launches as a public chatbot, followed by ChatGLM2/3 and CogVLM-17B, Zhipu’s first vision-language model.

Jun 2024

GLM-4 extends context to 128K tokens and adds multimodal understanding plus “All Tools” function calling.

Jan 2025

The US Commerce Department adds Zhipu AI to its Entity List; the company disputes the decision and keeps shipping – training and serving on Huawei and Cambricon chips.

Jul 2025

Zhipu rebrands internationally as Z.AI alongside GLM-4.5 (355B/32B active) – the first GLM with one-click third-party coding-agent compatibility.

Sep–Dec 2025

GLM-4.6 raises context to 200K; AutoGLM is open-sourced in December as the first stable phone-use agent; GLM-4.7 closes out the year focused on end-to-end coding.

Jan 2026

Z.AI lists on the Hong Kong Stock Exchange on January 8 – the world’s first publicly listed large-language-model company.

Feb–Apr 2026

GLM-5 launches February 11 (744B/40B-active MoE, 28.5T tokens); GLM-5.1 ships, trained entirely on non-Nvidia hardware and closing in on Claude Opus-class coding scores.

Jun–Jul 2026

GLM-5.2 (June 13) adds a 1M-token context window; ZCode launches July 2 as the agentic IDE – then, in July, gets self-hosted by Hugging Face to help contain an autonomous attack by escaped third-party models.

Aug 2026

GLM-5.3 launches August 14 with major coding and cybersecurity gains from post-training alone; GLM-5.3-Flash follows August 26 as the first natively multimodal GLM-5 model; GLM-5.3’s own weights arrive August 28 after a safety review.

07 · Staying productive

How to use GLM well

Be specific

State audience, goal, format, and constraints. “Rewrite this changelog for a non-technical customer, five bullets max” beats “clean this up.”

Match effort to the task

Routine edits don’t need max effort – save it for genuinely hard, long-horizon coding or security work, which is where GLM-5.3 was built to shine.

Choose your access path deliberately

The Coding Plan beats metered API pricing for steady daily agent use; the API suits bursty workloads; self-hosted weights suit regulated or sensitive data.

The prompt pattern that works

Role:You are a senior backend engineer reviewing a pull request for security issues.Task:Audit this diff for injection, auth-bypass, and unsafe deserialization risks.Format:A numbered list, severity-ranked, with a one-line fix suggested for each.

Role, task, format – with GLM-5.3’s coding and security strengths, a structured review prompt like this is where it earns its price. For long-running agent work, keep the system prompt stable so caching keeps costs down.

08 · The window

How large is the context window?

GLM-5.3 and GLM-5.3-Flash both carry a standard 1,048,576-token context window – roughly 750,000 words – with up to 128,000 tokens of output. GLM-4.6 and GLM-4.7 support 200,000 tokens (about 150,000 words); GLM-4.5 and the original 2023 ChatGLM topped out at 128,000 and 8,000 tokens respectively.

GLM-5.3 / GLM-5.3-Flash – 1,048,576 tokens≈ 750,000 words
GLM-4.6 / GLM-4.7 – 200,000 tokens≈ 150,000 words
GLM-4.5 (2025 launch) – 128,000 tokens≈ 96,000 words
ChatGLM (2023 launch) – 8,000 tokens≈ 6,000 words

Practical rule of thumb: a token is roughly 0.75 English words, and about 1.5 tokens per word once you factor in punctuation and spacing. As with any 1M-token model, attention is not perfectly uniform across the full window – expect some “lost in the middle” degradation on very long documents.

09 · Plans and pricing

What does GLM cost?

PlanPriceModel accessLimits
Z.AI chat$0GLM-5.3 chat, web searchUsage caps; no coding-agent tooling
Coding Plan – Lite$18/moGLM-5.3, GLM-5.3-FlashPoints quota; off-peak calls cost roughly half
Coding Plan – Pro$80/moGLM-5.3, GLM-5.3-FlashLarger allowances, priority access
Coding Plan – Max$168/moGLM-5.3, GLM-5.3-Flash, peak priorityHighest quota, team features
API – GLM-5.3$1.40 / $4.40per 1M tokens in/out, 1M context$0.26 cached input · text only
API – GLM-5.3-Flash$0.15 / $0.50per 1M tokens in/out, 1M context$0.03 cached input · multimodal
API – GLM-4.7-FlashFree200K contextFree tier, lower priority
Self-hostedFree (license varies)Full weights, unlimited useGLM-5.3-Flash is plain MIT; GLM-5.3 needs a security review above $10B revenue

Off-peak Coding Plan usage deducts roughly half as many credits per call, so weekend and evening agent runs go further. Calls to legacy GLM-5.2, GLM-5.1, and GLM-4.7 on the Coding Plan are automatically routed to GLM-5.3 or Flash. The chart above reflects mid-September 2026 – check Z.AI’s official pricing page for current rates.

10 · FAQ

Frequently asked questions

Is GLM free?

Yes. The Z.AI chat gives free access to GLM-5.3 with usage caps, and several API models (GLM-4.7-Flash, GLM-4.5-Air) are free or near-free. Meaningful daily coding-agent use starts around $18/month on the GLM Coding Plan.

Is GLM really open source?

Mostly, with one exception. GLM-5.3-Flash and most of the GLM-4 line ship under the permissive, unconditional MIT license. GLM-5.3 itself uses a custom license that’s MIT-style for nearly everyone, but requires large model-as-a-service operators (above $10B in trailing revenue) to pass a Z.AI security review first.

How does GLM compare to ChatGPT, Claude, or Gemini?

On coding and agentic benchmarks, GLM-5.3 is genuinely frontier-adjacent among open-weight models – at a fraction of the API price of Opus-class or GPT-5.6-class flagships. It ships without a published SWE-bench Verified score, and its flagship form is text-only (vision lives in Flash). Closed rivals still lead on ecosystem breadth and general-purpose polish outside coding.

Is my data safe with GLM?

Z.AI is a Chinese company on the US Entity List, and its hosted app and API process data outside most Western regulatory jurisdictions. For sensitive or regulated work, either avoid the hosted services or self-host the open weights and keep everything on infrastructure you control.

What is AutoGLM?

Z.AI’s phone- and computer-use agent, open-sourced in December 2025. It reads what’s on screen and simulates taps, swipes, and typing to complete multi-step tasks – ordering food, booking travel, drafting documents – with both on-device and cloud execution modes.

What are GLM’s limitations?

The GLM-5.3 flagship is text-only – vision lives in the separate GLM-5.3-Flash and the older GLM-4.5V/4.6V models. There’s no published SWE-bench Verified score, which complicates head-to-head procurement comparisons. And GLM-5.3’s dual-use cybersecurity strength is real enough that Z.AI delayed its weights for a safety review.

Heads-up:GLM ships new versions on a roughly quarterly cadence, and pricing, licensing, and model IDs change accordingly. The details above reflect September 2026 (GLM-5.3 released Aug 14, weights Aug 28). Check Z.AI’s official docs and Hugging Face pages for current terms.

Related reading