Z.ai
GLM 5.3
z-ai/glm-5.3
A large reasoning model from Z.ai built for complex software engineering and long-horizon agent tasks. Text in, text out, 1M-token context, with weights on Hugging Face.
RoleThe open-weight agent
$0.070
In / M
$1.4
Out / M
1.0M
Context
- AA intelligence index
- 31.7
- Inputs
- text
- Outputs
- text
- Tools
- Listed
- Structured output
- Listed
- Reasoning
- Listed
- Weights
- Open · zai-org/GLM-5.3
- Listed
- Aug 18, 2026
01 · What it's good at
Use it for this.
Three jobs this model is built for, and the ones to give to something else.
01
Long-horizon coding agents on an open-weight model
02
Keeping your options open: weights are published (zai-org/GLM-5.3)
03
Text-only reasoning over very long inputs
Skip it for
- Images or screenshots. This listing is text-only; GLM 5.3 Flash takes images and video
02 · Example tasks
What you'd actually send it.
Real inputs, not a benchmark. Paste one into the snippet below to try it.
You send
Split the 3,000-line utils.ts into focused modules. Keep exports stable and update imports across the repo.
Why this model
Long-horizon software work is the stated focus.
03 · Your stack
Where it sits: Plan & reason.
Use it as the agent model when open weights matter: start on the hosted API, and keep the option to self-host or switch providers later without changing models.
Layer 01
Retrieve
Layer 02 · this model
Plan & reason
Layer 03
Do the volume
Layer 04
Gate & route
Layer 05
Draw, speak, listen
04 · Wire it up
One request. Paste and run.
Uses OpenRouter, so one key works for every model on this site. Set OPENROUTER_API_KEY and swap the prompt for one of the tasks above.
// OpenRouter speaks the OpenAI Chat Completions format.
const res = await fetch("https://openrouter.ai/api/v1/chat/completions", {
method: "POST",
headers: {
Authorization: `Bearer ${process.env.OPENROUTER_API_KEY}`,
"Content-Type": "application/json",
},
body: JSON.stringify({
"model": "z-ai/glm-5.3",
"messages": [
{
"role": "user",
"content": "Split utils.ts into focused modules and list the new files with their exports."
}
],
"reasoning": {
"effort": "high"
}
}),
});
const data = await res.json();
console.log(data.choices[0].message.content);
05 · Watch out for
Gotchas.
Default reasoning effort on the listing is max. Lower it for routine turns.
Open weights does not mean cheap to host. The hosted API is usually the right place to start.
Sources
Lessons are written from each model's listing and the linked docs. We don't publish our own benchmark numbers; the index above is Artificial Analysis, as shown on OpenRouter.
Keep learning