Blog · GLM

GLM API Pricing (2026): Zhipu GLM-5.2 Cost & Access Guide

Zhipu GLM-5.2 is the budget end of the Chinese-model range. This guide covers what it costs, how to access it from outside China, and where it fits in a multi-model stack.

Last updated: 2026-10-08. All prices below are captured as of the dates noted — vendor pricing changes often, always verify on the official pricing page before buying.

1. What is GLM?

GLM (General Language Model) is developed by Zhipu AI, a Beijing-based AI company. The current generation is GLM-5.2, a cost-efficient model family positioned below DeepSeek Flash in reasoning capability and well below Qwen Max in breadth, but with the lowest per-token cost among the three major Chinese model families.

GLM is a practical choice for:

High-volume classification and extraction — tasks where unit cost is the primary constraint
Batch processing — large offline jobs where throughput matters more than per-request quality
Chinese-language NLP — segmentation, entity extraction, sentiment analysis on Chinese text
Fallback tier — keep GLM as the cheap path and escalate to pro or vip-glm when quality demands it

Unlike DeepSeek, GLM does not use peak/off-peak pricing. It also has no request-size tiering — a 200-token call costs the same per-token as a 4K-token call on the official platform. AI24X mirrors this with a flat rate across the board.

2. Official GLM pricing (Zhipu first-party)

Zhipu AI publishes its pricing on the official platform (bigmodel.cn). The following rates were captured from the official documentation (verified 2026-09-10):

Model	Input / 1M tokens	Output / 1M tokens	Context
`GLM-5.2-Flash`	¥0.60 (≈ $0.08)	¥2.00 (≈ $0.28)	128K
`GLM-5.2-Pro`	¥5.00 (≈ $0.70)	¥14.00 (≈ $1.95)	128K
`GLM-5.2-Plus`	¥10.00 (≈ $1.40)	¥30.00 (≈ $4.18)	128K

Official source: https://bigmodel.cn/pricing (in CNY). USD approximations use 2026-09-10 exchange rate ~7.18 CNY/USD.

Three things to note about the first-party platform:

Chinese phone required for registration — international numbers are not accepted.
Chinese payment methods only — Alipay, WeChat Pay, or China-issued UnionPay cards. No PayPal, no international Visa/Mastercard.
No peak/off-peak surcharge — the price is the same regardless of when you call. This is simpler than DeepSeek's two-tier system but means you cannot save by shifting your schedule.

For international developers, the first-party platform is effectively inaccessible unless you have Chinese credentials. That is the gap third-party gateways fill.

⚠️ Clarification: This article is written by AI24X, an AI gateway — we are not the official Zhipu AI service and we are not affiliated with Zhipu. We explain the official pricing for context and offer our own access as one of several alternatives.

3. GLM vs DeepSeek vs Qwen — pricing comparison

How does GLM compare to the other major Chinese model families? Here is a side-by-side of the cheapest tier from each (official first-party rates, captured 2026-09-10):

Model	Cheapest tier	Input / 1M	Output / 1M	Peak surcharge?
GLM-5.2-Flash (Zhipu)	Flash	¥0.60 ($0.08)	¥2.00 ($0.28)	No
DeepSeek V4.1 Flash (DeepSeek)	Flash off-peak	$0.15	$0.60	Yes (2× peak)
Qwen Max (Alibaba)	Max	¥2.00 ($0.28)	¥6.00 ($0.84)	No

GLM-5.2-Flash is the cheapest per-token of the three on the official platform — roughly half the input cost of DeepSeek Flash off-peak, and a third of Qwen Max. The trade-off is that GLM-5.2 is in a lower capability tier: it handles classification, extraction, and simple generation well but falls behind on complex reasoning, multi-step instructions, and creative writing.

Through an aggregator like AI24X, the comparison shifts because you pay a flat rate and the gateway handles routing. AI24X flash tier starts at $0.35/M output (verified 2026-10-02) and covers DeepSeek, Qwen, GLM, Kimi, and MiniMax under one key — so you do not pick a single "cheapest" model; the gateway picks the best channel per request.

4. Accessing GLM from outside China

If you are outside China and do not have a Chinese phone number, you have two paths to GLM access:

Path A: AI gateway (recommended for most developers)

AI gateways like AI24X provide GLM access through an OpenAI-compatible endpoint. No Chinese phone number needed — signup uses email, payment uses PayPal or international card.

Through AI24X, GLM is available under the vip-glm model ID. The same key also works for DeepSeek, Qwen, Kimi, MiniMax, MiMo, GPT-5/6, Claude, and Gemini — all through one POST /v1/chat/completions endpoint. Smart routing picks the best upstream channel per request, with automatic model switching.

Path B: Direct Zhipu platform (requires Chinese credentials)

If you have a Chinese phone number and a Chinese payment method, you can register at bigmodel.cn and get an API key directly. The endpoint is https://open.bigmodel.cn/api/paas/v4 and requires a Zhipu-specific SDK or a manually constructed JWT for authentication (it does not use a simple Bearer token).

5. One key for GLM + DeepSeek + Qwen + Kimi

The practical advantage of using an aggregator instead of managing separate accounts:

One OpenAI-compatible endpoint: POST /v1/chat/completions with Authorization: Bearer <key>
One payment rail: PayPal or card — no Chinese phone number required
All Chinese model families on one key: GLM, DeepSeek, Qwen, Kimi, MiniMax, MiMo — plus GPT-5/Claude/Gemini when a task needs flagship power
Flat-rate billing: no peak/off-peak tracking, no request-size tiers, no per-model minimums
Smart routing: the gateway evaluates available upstream channels and picks the best one by price, speed, and stability, with automatic model switching when a channel degrades

6. Worked cost example

Say a developer sends 50M input tokens and 10M output tokens per month to a budget-tier Chinese model:

Scenario	Input bill	Output bill	Total
GLM-5.2-Flash official (CNY)	50 × ¥0.60 = ¥30.00	10 × ¥2.00 = ¥20.00	¥50.00 (~$6.96)
DeepSeek V4.1 Flash official (all off-peak)	50 × $0.15 = $7.50	10 × $0.60 = $6.00	$13.50
DeepSeek V4.1 Flash official (all peak)	50 × $0.30 = $15.00	10 × $1.20 = $12.00	$27.00
AI24X flash tier (flat, multi-model)	— (flat output pricing)	~10 × $0.35 = $3.50	varies by routing — often the effective bill lands below official peak, with no scheduling overhead

The point is not that any single row always wins — it is that your real cost depends on your call pattern and region. If you are outside China and cannot use the official CNY platform, the relevant comparison is between aggregator flat rates. AI24X flash from $0.35/M output covers GLM, DeepSeek, Qwen, and others — the value is not "cheapest per token on one model" but "one predictable bill across the whole model stack."

7. Copy-paste setup

Base URL (OpenAI-compatible)

https://api.ai24x.com/v1

Quick test with curl

curl -X POST "https://api.ai24x.com/v1/chat/completions" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer ***" \
  -d '{"model":"vip-glm","messages":[{"role":"user","content":"What is GLM-5.2?"}]}'

Python with OpenAI SDK

from openai import OpenAI

client = OpenAI(
    base_url="https://api.ai24x.com/v1",
    api_key="sk-..."
)

response = client.chat.completions.create(
    model="vip-glm",
    messages=[{"role": "user", "content": "Summarize this document."}]
)
print(response.choices[0].message.content)

Existing OpenAI SDKs and tools (Open WebUI, LobeChat, Cursor, Dify, LiteLLM, n8n) connect the same way: change base URL + key, keep the rest of your code.

8. FAQ

Is GLM free?

Zhipu offers a free tier on the official platform (limited daily quota for the cheapest model). Through AI24X, a shared free channel of 100K tokens/day is available when your balance is empty — see the console for current policy.

Can I use the GLM API without a Chinese phone number?

Not on Zhipu's first-party platform. Through AI24X, yes — signup requires only an email address, and payment is by PayPal or international card. See the setup section above.

Does GLM have peak/off-peak pricing?

No. Zhipu charges a flat rate regardless of call time. AI24X also bills flat — no peak surcharge, no request-size tiering.

Is AI24X the official GLM API?

No. AI24X is an independent AI gateway (not affiliated with Zhipu AI) that routes to GLM and other models through one OpenAI-compatible key.

What context length does GLM-5.2 support?

GLM-5.2 supports up to 128K tokens of context across all tiers — sufficient for most document-length tasks. For 1M-context workloads, DeepSeek or Kimi are better choices.

What is the refund policy?

Refund rules are set per platform. AI24X's billing is prepaid-credit based; check the console's billing/refund terms for the current policy, and top up only what you plan to use.

9. Getting started

Register at https://www.ai24x.com and create an API key in the console.
Top up at console → billing (PayPal or card).
Use the base_url + key from the setup section in any OpenAI-compatible tool — done.

Developer docs: https://open.ai24x.com · Console: https://www.ai24x.com/console.html · Pricing: https://www.ai24x.com/pricing.html · Model catalog: GLM on AI24X

Educational/informational article. This page is not affiliated with Zhipu AI and does not constitute investment advice of any kind. All prices are as captured on the dates noted and may change; always check official pricing pages.

One API. Every AI. Pay Less.

Start Building Free Get API Key View pricing