Production-grade GLM-5.2 inference for developers — an OpenAI-compatible API priced at 50% of the Zhipu US list rate. Same model, lower bill.
We host GLM-5.2 ourselves on dedicated hardware, so we pass the savings straight to you.
Every token and every plan is priced at 50% of the equivalent Zhipu US list rate. No usage tiers to game, no fine print.
OpenAI-compatible endpoints. Change the base URL and API key — your existing SDKs, agents, and tools just work.
Dedicated GPU capacity hosted in Southeast Asia (Malaysia), tuned for steady throughput on long-running agent workloads.
Streaming, function calling, and long context — the full GLM-5.2 capability set, ready for real applications.
A clean dashboard for keys, spend, and rate limits. Know exactly what you're using and what it costs.
Billing in USD/EUR, English support, and terms written for developers shipping to Western markets.
Pay per token, or subscribe monthly — always exactly 50% of the equivalent Zhipu US (z.ai) list price.
Use GLM-5.2 inside Claude Code, Cline, Roo Code, Kilo Code and other OpenAI-compatible agents. One monthly fee, a generous prompt quota — no per-token metering.
Prices reflect Zhipu's published US (z.ai) list rates as of July 2026, halved. GLM-5.2 API list price is $1.40 input / $4.40 output per 1M tokens; the Coding Plan tiers are $18 / $72 / $160 per month. We peg to Zhipu's list price and charge exactly 50%. Join the waitlist and we'll send the confirmed, final price list before you're ever asked to pay.
from openai import OpenAI client = OpenAI( base_url="https://api.halfstack.dev/v1", api_key="sk-your-key", ) resp = client.chat.completions.create( model="glm-5.2", messages=[ {"role": "user", "content": "Ship it."}, ], stream=True, ) # same SDK. same code. half the cost.
Yes. We run the GLM-5.2 model on our own dedicated GPU infrastructure. You get the full model — the difference is our cost structure, not a smaller or distilled version.
We self-host on dedicated hardware in Southeast Asia and run lean. That lets us peg every rate to Zhipu's published US list price and cut it in half.
Use the pay-per-token API for production apps and scripts — you pay only for what you use, with no tool restrictions. Use the Coding Plan if you live inside a coding agent (Claude Code, Cline, Roo Code…) all day — a flat monthly fee caps your bill. Both are exactly 50% of Zhipu's US price.
We're finalizing deployment now. Join the waitlist and you'll be among the first emailed when API keys open up — waitlist members get first access.
Our inference runs in Malaysia on dedicated capacity. It's well-suited to agent and batch workloads; we'll publish real latency benchmarks for US/EU regions before launch.
If you already use the OpenAI SDK (or any OpenAI-compatible client), you only change the base URL and API key. No code rewrite.
No. Your prompts and completions are not used for training. Full data-handling terms will be published before launch.
Capacity is limited at launch, and we onboard in waitlist order. The earlier you join, the sooner you get keys and the confirmed price list.