GLM-5.3 Prime
z-ai/glm-5.3-prime
Guide only
- Released
- Sep 23, 2026
- Context window
- 1M
- Max output
- 131K
- In / out per 1M
- $2.80 / $8.80
- Inputs
- Text
GLM-5 free online chat · no sign-up
Talk to GLM-5.3, free, right here.
glm5.org is an independent guide to Z.ai's GLM-5 model family, with a chat that runs the real models. Ask GLM-5.3 Flash anything — or switch to GLM-5.3 for hard reasoning — without an account, a card or an install.
- Context
- 1M
- In chat
- 2
- Open repos
- 5
- Price here
- $0
attention · causal mask
§ 01The lineup
Every GLM-5 model, on one sheet
From the original open-weights GLM-5 to the multimodal GLM-5.3 Flash. Two of them run in our chat; the rest are documented with prices, context and licences.
GLM-5.3 FlashX
z-ai/glm-5.3-flashx
Guide only
- Released
- Sep 18, 2026
- Context window
- 1.05M
- Max output
- 131K
- In / out per 1M
- $0.37 / $1.25
- Inputs
- Text · Image · Video
GLM-5.3 Flash
z-ai/glm-5.3-flash
In our chatOpen weights
- Released
- Aug 26, 2026
- Context window
- 1.05M
- Max output
- 944K
- In / out per 1M
- $0.0052 / $0.1985
- Inputs
- Text · Image · Video
GLM-5.3
z-ai/glm-5.3
In our chatOpen weights
- Released
- Aug 18, 2026
- Context window
- 1.05M
- Max output
- 944K
- In / out per 1M
- $0.079 / $1.40
- Inputs
- Text
GLM-5.2
z-ai/glm-5.2
Guide onlyOpen weights
- Released
- Jun 16, 2026
- Context window
- 1.05M
- Max output
- 944K
- In / out per 1M
- $0.0341 / $7
- Inputs
- Text
GLM-5.1
z-ai/glm-5.1
Guide onlyOpen weights
- Released
- Apr 7, 2026
- Context window
- 205K
- Max output
- 131K
- In / out per 1M
- $1.40 / $4.40
- Inputs
- Text
GLM-5V-Turbo
z-ai/glm-5v-turbo
Guide only
- Released
- Apr 1, 2026
- Context window
- 203K
- Max output
- 131K
- In / out per 1M
- $1.20 / $4
- Inputs
- Text · Image · Video
GLM-5-Turbo
z-ai/glm-5-turbo
Guide only
- Released
- Mar 15, 2026
- Context window
- 203K
- Max output
- 131K
- In / out per 1M
- $1.20 / $4
- Inputs
- Text
GLM-5
z-ai/glm-5
Guide onlyOpen weights
- Released
- Feb 11, 2026
- Context window
- 205K
- Max output
- 131K
- In / out per 1M
- $1 / $3.20
- Inputs
- Text
Prices are per 1 million tokens on OpenRouter at the time of writing; they can change. Release dates are the dates the models were listed there.
§ 02Why GLM-5
What the GLM-5 family is good at
Z.ai builds GLM-5 for long, practical work: shipping code, running agent loops and reading very long inputs.
Coding and agentic engineering
The family is aimed at real software work rather than chat tricks. Z.ai positions GLM-5.1 and GLM-5.3 for long-horizon coding: planning a change, editing many files, running tools and fixing what breaks.
A million tokens of context
GLM-5.2, GLM-5.3 and GLM-5.3 Flash accept about one million tokens in a single request. That is enough for a large codebase, a long contract bundle or a book-length transcript.
Open weights you can download
Five GLM-5 models are published on Hugging Face under the zai-org organisation, most of them under the MIT licence. You can study them, fine-tune them and run them on your own hardware.
Images and video with Flash
GLM-5.3 Flash is natively multimodal: it reads images and video as well as text. In our chat you can attach a photo, a chart or a screenshot and ask about it.
Adjustable reasoning
GLM-5.2 and later expose reasoning effort levels, so you can trade speed for care. Our Think deeper switch maps to exactly that control.
Low prices for the capability
On OpenRouter, GLM-5.3 Flash costs a fraction of a cent per million input tokens and under $0.20 per million output tokens, which is why we can offer it free.
§ 03How to use it
From question to answer in four steps
The chat works the same on a phone and on a desktop. Nothing to install, nothing to sign.
Open the chat
Type into the box above or open the chat page. GLM-5.3 Flash is selected by default and answers in seconds.
Pick a model
Stay on GLM-5.3 Flash for everyday work and image questions. Switch to GLM-5.3 when a task needs heavier reasoning, such as tricky code or a long analysis.
Add context
Paste the code, the text or the data you are working on, or attach one image. Say who the answer is for and what format you want: a table, a list, a short email.
Refine
Ask follow-up questions, edit your last message, regenerate an answer, or turn on Think deeper. Your chats stay in your browser so you can come back to them.
§ 04Open weights
Most of GLM-5 is downloadable
Unlike most frontier model families, a large part of GLM-5 is published as open weights. GLM-5, GLM-5.1, GLM-5.2 and GLM-5.3 Flash are released under the MIT licence; GLM-5.3 uses its own custom licence, so read its terms before you build on it.
These are big models — GLM-5 has 744 billion parameters in total, with 40 billion active per token — so running them at full precision needs a multi-GPU server. The official model cards show how to serve them with vLLM and SGLang.
- zai-org/GLM-5MIT
- zai-org/GLM-5.1MIT
- zai-org/GLM-5.2MIT
- zai-org/GLM-5.3GLM-5.3 (custom)
- zai-org/GLM-5.3-FlashMIT
§ 05API
Call GLM-5 from your own code
Every GLM-5 model speaks the OpenAI-compatible chat completions format. You can use it through Z.ai's own platform or through OpenRouter with a single key for many providers.
Switching an existing app is usually a matter of changing the base URL and the model name. The snippet below asks GLM-5.3 Flash a question through OpenRouter.
curl https://openrouter.ai/api/v1/chat/completions \
-H "Authorization: Bearer $OPENROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "z-ai/glm-5.3-flash",
"messages": [{"role": "user", "content": "Hello, GLM-5!"}]
}'§ 06Comparisons
GLM-5 against the field
Context, output limits, inputs and prices side by side with DeepSeek, Qwen, Claude, GPT-6 and Gemini. Numbers come from the public OpenRouter model list, not from marketing slides.
| Model | GLM-5.3 | GLM-5.3 Flash | DeepSeek V4.1 Flash | Qwen3.7 Plus | Claude Sonnet 5.5 | GPT-6 Sol | Gemini 3.8 Flash |
|---|---|---|---|---|---|---|---|
| Developer | Z.ai | Z.ai | DeepSeek | Qwen | Anthropic | OpenAI | |
| Released | Aug 18, 2026 | Aug 26, 2026 | Sep 10, 2026 | Jun 3, 2026 | Sep 28, 2026 | Sep 22, 2026 | Sep 2, 2026 |
| Context window | 1.05M | 1.05M | 1.05M | 1M | 1M | 1.05M | 1.05M |
| Max output | 944K | 944K | 944K | 131K | 128K | 128K | 66K |
| Input price | $0.079 | $0.0052 | $0.3 | $0.32 | $2 | $2 | $0.825 |
| Output price | $1.40 | $0.1985 | $1.20 | $1.28 | $10 | $10 | $4.13 |
| Inputs | Text | Text, Image, Video | Text, Image | Text, Image | Text, Image, Files | Text, Image, Files | Text, Image, Video, Audio, Files |
| Open weights | Yes (Hugging Face) | Yes (Hugging Face) | Yes (Hugging Face) | No | No | No | No |
| Free chat here | In our chat | In our chat | — | — | — | — | — |
Prices are USD per 1 million tokens as listed by OpenRouter; the best value in each row is highlighted. Source: OpenRouter model list and the official Hugging Face model cards.
§ 07Use cases
What people use GLM-5 for
Practical guides with copy-ready prompts, the right model for each job and honest notes on limits.
7.1
Coding
Debug, review and refactor with the family's strongest skill.
7.2
Writing
Drafts, rewrites and edits in the tone you need.
7.3
Translation
Natural translations that keep tone, terms and format.
7.4
Study
Explanations, practice questions and study plans.
7.5
Data analysis
SQL, pandas and plain-language insights from your data.
7.6
Agents
Plan and build tool-using agents on GLM-5.
§ 08FAQ
GLM-5 questions, answered
Short, sourced answers about the models and about this site.
Q01What is GLM-5?
GLM-5 is a family of large language models made by Z.ai (Zhipu AI). The original GLM-5 arrived in February 2026 and was followed by GLM-5.1, GLM-5.2, GLM-5.3 and several fast or multimodal variants such as GLM-5.3 Flash and GLM-5 Turbo.
Q02Can I use GLM-5 for free?
Yes. On glm5.org you can chat with GLM-5.3 Flash and GLM-5.3 for free, without signing up. There are fair daily limits per visitor, and GLM-5.3 has a smaller daily quota than Flash. Developers can also download the open-weights models from Hugging Face.
Q03Which GLM-5 model should I pick?
Start with GLM-5.3 Flash: it is fast, inexpensive, reads images and handles most everyday tasks well. Move to GLM-5.3 for complex coding, long multi-step reasoning or anything where quality matters more than speed.
Q04Is GLM-5 open source?
Several GLM-5 models are open weights. GLM-5, GLM-5.1, GLM-5.2 and GLM-5.3 Flash are published on Hugging Face under the MIT licence, and GLM-5.3 is published under a custom GLM-5.3 licence. GLM-5 Turbo, GLM-5V-Turbo, GLM-5.3 FlashX and GLM-5.3 Prime are available only through APIs.
Q05How big is the GLM-5 context window?
The original GLM-5 accepts about 205K tokens on OpenRouter. GLM-5.2, GLM-5.3 and GLM-5.3 Flash raise that to about 1.05 million tokens, and GLM-5.3 Prime lists a 1 million-token window.
Q06Is glm5.org the official GLM website?
No. glm5.org is an independent project. It is not affiliated with or endorsed by Z.ai or Zhipu AI. The chat uses the official GLM models through OpenRouter, and the guides are written by us from public sources such as the Hugging Face model cards and Z.ai's documentation.
Q07Does the chat keep my conversations?
Your conversations are saved only in your own browser so that you can return to them. Our server passes your messages to the model and returns the answer, but it does not store the content of your chats.
Try it now
Ask GLM-5.3 something real.
No account, no card, no install. Type a question and watch GLM-5.3 Flash answer in seconds — switch to GLM-5.3 for the hardest problems.
Open the free chatFree daily limits apply so the chat stays open for everyone.