glm-5.2

Zhipu / Z.ai

This is the exact string to pass as model in a request. IDs are case-sensitive, and this one is billed per token.

Prices (USD per 1M tokens)

Input

$0.6200

Output

$3.10

Cache read

$0.0620

Cache write

$0.6200

Billing reads the same table this page does, so the price you see is the price you are charged. Cache read/write apply when the upstream provider reports cached tokens.

Below the vendor's list price

Input: official $1.40 per 1M → ours $0.6200 (55% less)

Output: official $4.40 per 1M → ours $3.10 (29% less)

Vendor list price from the vendor's own pricing page, checked 2026-09-23. Vendors change prices without notice, and some run time-limited discounts or batch tiers — if the vendor's page differs today, their page is the truth.

Specs

Provider: Zhipu / Z.ai
Context window: 1M
Image input: Supported
Best for: Everyday chatCodingAgentsLong documentsImage understanding
Key required: Any Standard key. No dedicated key needed.
Data: Request content is not stored here. It is forwarded to the upstream provider of this model, which handles it under its own policy — privacy policy.

How to call it

The OpenAI-compatible chat endpoint, POST https://jzstoken.com/api/v1/chat/completions. This is the format the client guides assume and the one our own catalog listings are checked against before they can be called at all.

curl https://jzstoken.com/api/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-YOUR_KEY_HERE" \
  -d '{
    "model": "glm-5.2",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Replace sk-YOUR_KEY_HERE with a key from your dashboard. Top-ups start at $1.50; see Quick Start for the full walkthrough and the Python SDK version.

Clients that require the Anthropic Messages format can use POST https://jzstoken.com/api/v1/messages instead — see the Claude Code and Cursor guides. Which of these two formats any individual model supports through our gateway has not been certified per model, so try the chat endpoint first.

Not listed here yet

We would rather leave these blank than fill them with a number nobody verified: maximum output tokens, per-model rate limits, and per-model certification for streaming, tool calling or structured output. None of that is returned by the upstream pricing API and we have not measured it. If you need a specific figure for capacity planning, ask us and we will measure it rather than estimate.