New from Z.ai

GLM-5.3 online: Z.ai's open-weight coding model, pay per use

Z.ai released GLM-5.3 on August 14, 2026: a reasoning model for complex software engineering and long agent runs, with about a million tokens of context and open weights published two weeks later. It shares the base of GLM-5.2 and gains everything from post-training, lifting Terminal-Bench 3.0 from 4.6 to 28.3 percent and topping CyberGym. On avots.ai you use GLM-5.3 and the fast multimodal GLM-5.3 Flash pay per use, no plan, and the same balance also covers GPT-6 Astra, Claude Fable 5.1, Gemini and 40+ image, video and music models.

Use GLM-5.3 now ✦ See pricing
Two tiers, one predecessor

GLM-5.3, GLM-5.3 Flash and GLM-5.2 side by side

 GLM-5.3GLM-5.3 FlashGLM-5.2
Best forterminal coding, security, long agentsvolume work, images and video inthe previous generation
Price per 1M in / out$1.40 / $4.40$0.07 / $0.25$1.40 / $4.40
Context window1M tokens1M tokens1M tokens
Inputtexttext, images, videotext
Terminal-Bench 3.028.3%4.6%
On avots.aipay per usepay per usepay per use

Same base as GLM-5.2, same price, six times the Terminal-Bench score: 5.3 simply replaces 5.2 for coding. Flash is the multimodal budget tier of the same generation at a twentieth of the price. Switch freely, one balance covers all three.

What is new

What GLM-5.3 brings

Top of CyberGym

84.5 percent on the cybersecurity benchmark, which Z.ai reports as ahead of Claude Mythos 5 and GPT-5.6 Sol.

Six times better in the terminal

Terminal-Bench 3.0 rose from 4.6 to 28.3 percent through post-training alone, on the same base as GLM-5.2.

Open weights

Z.ai published the full model on August 28 after two weeks of safety hardening. Hosted here, self-hostable if you have the GPUs.

Flash sees images and video

The Flash tier reads screenshots, documents and clips with a hybrid attention design that keeps long contexts accurate at a very low price.

Three steps

How to use GLM-5.3

Create an account

Sign up on app.avots.ai with email, Google or Telegram. Takes a minute.

Top up from 0.99 EUR

1,000 tokens cost 0.99 EUR. No plan to choose, no renewal, the balance sits there until you use it.

Pick GLM-5.3 or Flash

Choose a GLM-5.3 tier in the model picker, or let Agent Avots route the task for you. Every reply shows its token cost.

Sign up and start
One balance

Where you can use GLM-5.3

Web app

app.avots.ai on desktop and mobile, no install needed.

Telegram

GLM-5.3 in @AvotsAIbot, same balance, from any phone.

MCP

Use it inside Claude Desktop, Cursor or Cline via the MCP server.

Try GLM-5.3 for under a euro

1,000 tokens cost 0.99 EUR and a GLM-5.3 Flash reply costs a few of them. The same balance unlocks GPT-6 Astra, Claude Fable 5.1, Gemini, Veo video and Nano Banana images.

Sign up and chat ✦
Guides

Related pages

GLM-5.2

The previous Z.ai generation, same price, still in the picker.

DeepSeek V4

The other open-weight heavyweight, 1M context, top coding scores.

Kimi K3

Moonshot's frontier open model, number one at coding, 1M context.

GPT-6 Astra

OpenAI's new flagship and its Pro mode, pay per use.

avots.ai runs 40+ AI models for chat, image, video and audio on one balance. See the full list on the Tools page or open the app and start creating.

Browse all model pages Open the app
FAQ

GLM-5.3 questions

What is GLM-5.3?
GLM-5.3 is Z.ai's reasoning model released on August 14, 2026, built for complex software engineering and long-horizon agent tasks. It shares the base of GLM-5.2 and gains everything from post-training: Terminal-Bench 3.0 went from 4.6 to 28.3 percent. It has about a 1 million token context window, and the full weights were published on August 28.
What is GLM-5.3 Flash?
The fast multimodal tier of the same generation: it reads text, images and video, keeps the long context, and costs 0.07 USD per million input tokens and 0.25 per million output. Use it for high-volume coding help and agent loops where the big model is not needed.
How much does GLM-5.3 cost on avots.ai?
Pay per use, no subscription. GLM-5.3 is priced at 1.40 USD per million input tokens and 4.40 USD per million output; GLM-5.3 Flash at 0.07 and 0.25. On avots.ai you top up from 0.99 EUR and every reply shows its token cost before you run it.
What is GLM-5.3 best at?
Software engineering in a terminal and security work. Z.ai reports 84.5 percent on CyberGym, ahead of Claude Mythos 5 and GPT-5.6 Sol on that benchmark, and a six-fold jump on Terminal-Bench 3.0 over GLM-5.2. For everyday chat it is a capable, inexpensive model with a very long context.
Can I run GLM-5.3 myself instead?
The weights are open, but the full model needs a rack of GPUs. On avots.ai it runs hosted, pay per use, in the same picker as GPT-6 Astra, Claude Fable 5.1 and Gemini, so you can compare them on one task without setting anything up.
Try it on avots ✦
Start creating