GLM-5.3-Flash
glm-5.3-flash
🇨🇭 Switzerland
GLM-5.3-Flash di Z.ai (licenza MIT), la versione più leggera della famiglia GLM-5.3: una mistura di esperti da 320 miliardi di parametri, circa 18 per token. Forte con il codice e con gli agenti che usano strumenti. Ragiona sempre prima di rispondere e il ragionamento conta come token in uscita: con reasoning_effort "low" ragiona poco e risponde in pochi secondi, con "high" o senza indicazioni ragiona di più. Contesto di 262.144 token, pesi a 4 bit (NVFP4) pubblicati da NVIDIA. Gira su NVIDIA RTX PRO 6000 Blackwell, in Svizzera. In prova.
At a glance
| Publisher | Z.ai |
|---|---|
| Architecture | Mixture of experts, 320 billion parameters, about 18 billion per token |
| Licence | MIT |
| Precision | NVFP4, 4 bits (NVIDIA) |
| Context | 262,144 tokens, prompt and answer together |
| Longest answer | max_tokens up to 32,768, reasoning included |
| Reads | Text |
| Calls tools | Yes |
| JSON schema | Yes |
| Reasons | Yes, always, at three levels: reasoning_effort "low", "medium" or "high". The reasoning counts as output tokens. |
| Runs in | 🇨🇭 Switzerland |
| Status | Loaded: the first request answers without a start-up wait. Live state on Status. |
| Price | glm-5.3-flash@blackwell6000pro: € 0,230000 input · € 0,800000 output , per million tokens, excluding VAT. All prices |
Use it
curl https://api.ai-tokens.it/v1/chat/completions \
-H "Authorization: Bearer $API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "glm-5.3-flash",
"messages": [{"role": "user", "content": "Summarise in one line: the meeting moves to Thursday at 10."}]
}'
import os
from openai import OpenAI
client = OpenAI(base_url="https://api.ai-tokens.it/v1", api_key=os.environ["API_KEY"])
answer = client.chat.completions.create(
model="glm-5.3-flash",
messages=[{"role": "user", "content": "Summarise in one line: the meeting moves to Thursday at 10."}],
)
print(answer.choices[0].message.content)
More in Quickstart and Chat completions.
Limits
- Prompt and answer together stay within 262,144 tokens; beyond that the engine refuses the request with a
400that says so. - The reasoning counts against
max_tokens: a low cap can end the answer before it starts, withfinish_reason: "length". - Text only: a request with an image gets a
400. - Served at 4 bits, which makes it faster and fits our machines; we write the precision of every model.
- A request goes only to the zones your key allows; if the model has no machine there now, a
503names the zones. It is never sent elsewhere.