Provider

Z.ai GLM API: models & pricing

GLM-5, GLM-5.1 and GLM-5.2 text models from Z.ai (Zhipu). Call every one with a single ATP project key through an OpenAI-compatible gateway.

Models
3
Z.ai
Modalities
Text
From
$3.20
per 1M output tokens
glm-5
Latest release
GLM-5.2
Jun 16, 2026
glm-5-2
Models

All Z.ai models on ATP

Prices in USD. Credits are deducted only when a request succeeds. Open a model for full pricing and an estimate.

Prices

3
USD · USD 1 = 100 credits
All Z.ai models on ATP
ModelOutput / rate
glm-5-1
Positioning: Flagship. Price level 2 of 4 among text models.
$4.84in $1.54
glm-5-2
Positioning: Flagship. Price level 2 of 4 among text models.
$4.40in $1.40
glm-5
Positioning: Balanced. Price level 2 of 4 among text models.
$3.20in $1.00
On ATP

How teams use Z.ai on ATP

The same Z.ai models, with the controls a company needs around them: who can call which model, from which project, within what budget, and a record of every request.

  • Project API keys

    Every key belongs to one project and inherits its models and balance. Revoke one at any time without touching the rest.

  • Model allowlist per project

    Enable only the Z.ai models a project needs. Anything off the list can't be called with that project's keys.

  • Budgets

    Credits flow from organization to workspace to project. The allocation is the cap, so one team can't spend another team's budget.

  • Per-request logs

    Each call is recorded with its project, key, model, usage and credits debited, so spend is attributable down to a single request.

  • One wallet, one bill

    Z.ai and the other 10 providers on ATP draw from the same prepaid credits. No separate vendor contracts or invoices.

quickstart.py · OpenAI SDK
from openai import OpenAI

client = OpenAI(
  base_url="https://api.atptoken.ai/v1",
  api_key="atp-•••",
)

r = client.chat.completions.create(
  model="glm-5",
  messages=[{"role": "user", "content": "hi"}],
)
print(r.choices[0].message.content)

Point an existing SDK or HTTP client at ATP and set the model id. Confirm what your key can call with GET /v1/models. Full guide in the docs

Governance treeSample
Organization  Northwind Labs · 9,842 credits
Workspace  Product
  62%  support-bot
models: glm-5, gpt-5.5
  38%  web-app
models: gpt-5.5,
        deepseek-v4-flash
Workspace  Media lab
 100%  video-lab
limit reached
models: seedance-2-0,
        kling-v3-pro
FAQ

Z.ai GLM API on ATP: FAQ

FAQ

6
How do I call the Z.ai GLM API through ATP Token?
Create a project API key, enable the Z.ai model ids you need on that project's model allowlist, point an OpenAI-compatible SDK at https://api.atptoken.ai/v1 and set model to the API id (for example glm-5).
How much does the Z.ai GLM API cost on ATP Token?
Prices for the 3 Z.ai models on ATP Token: input $1.00–$1.54 and output $3.20–$4.84 per 1M tokens. Usage is billed in credits (USD 1 = 100 credits), debited from the project balance per completed request; failed requests are not billed. No monthly fee, USD 5 minimum top-up.
Which Z.ai model is cheapest on ATP Token?
By price: GLM-5 (glm-5) at $1.00 input / $3.20 output per 1M tokens.
What are the newest Z.ai models on ATP Token?
The most recent Z.ai release we list is GLM-5.2 (glm-5-2), released by Z.ai on Jun 16, 2026. GET /v1/models lists the full set.
Do Z.ai models support tool calling and vision?
Per the capability flags in the ATP model catalog: of 3 text models, 3 support tool calling, 1 accept image (vision) input (GLM-5 (glm-5)) and 3 support extended thinking. Each model's flags are in the table above.
How do teams control who can use Z.ai models, and how much they spend?
Each project has its own API keys and model allowlist — a Z.ai model that isn't on the list can't be called with that project's key. Credits flow from organization to workspace to project with budget caps, every request is logged with project, key, model and credits debited, and all providers share one wallet and one bill.
Get started

One key for Z.ai and every other model.

Sign up free, top up from USD 5, and scope your first project key in minutes.