GLM-5.2
zai-org/GLM-5.2
- Provider
- Z AI
- Input / 1M
- $1.40
- Output / 1M
- $4.40
- Cached / 1M
- $0.18
- Context
- 1M tokens
Flagship. Long-horizon agentic tasks. 744B-param MoE (40B active).
Sovereign AI: private AI infrastructure on hardware you rent or own, fully managed.
Pricing
Sovereign AI and Sovereign HPC, fully managed by national lab engineers. Start in Hoonify Cloud, then move to hardware you rent or own on one annual subscription.
How pricing works
Pick where Hoonify runs, and the pricing follows:
Pay as you go
Hoonify Cloud. AI is billed per 1M tokens. HPC is billed by usage. No minimums.
Flat subscription
Your hardware, on-prem or air-gapped. Priced per GPU or CPU node, with no meter.
Move when ready
Start in our cloud and move on-prem later. Your code and workflows stay the same.
Model pricing
zai-org/GLM-5.2
Flagship. Long-horizon agentic tasks. 744B-param MoE (40B active).
google/gemma-4-31B-it
Best value. Dense multimodal model for deep reasoning. 140 languages.
Qwen/Qwen3.6-27B
Vision. 27B dense model with native vision encoder. 201 languages.
All prices in USD per 1M tokens. Prices are live and may change with network capacity, the catalog is always the source of truth.
Cost savings calculator
Estimate what you would pay on Hoonify versus your current provider. Adjust the volume and your current rate — the savings update live.
Your current cost — per 1M tokens
Compare against a Hoonify model
Estimated monthly savings
$1,238
100.0× cheaperabout $14,850 a year
Estimate · blended in/out rate · taxes excluded

Migrate in 30 seconds
Hoonify is OpenAI-compatible. Point your existing SDK at our base URL and you're running on the operator network.
Sovereign
When your data can't leave the building, run Sovereign AI, Sovereign HPC, or both inside it. Our national lab engineers deploy and manage it, and you pay one annual subscription sized to your GPUs and CPU nodes, with no per-token billing.
Get Sovereign PricingAnnual subscription
The more you deploy, the less each one costs (multi-year saves more) = volume tiers
Volume tiers and multi-year discounts. Talk to us for your quote.
On-prem · Air-gapped
Zero data leaves your network (on-prem or air-gapped) = 0 bytes out
On your servers, in your private cloud, or fully air-gapped.
AI X HPC
Sovereign AI and Sovereign HPC (no second vendor) = 1 vendor
Start with either and add the other, with no second vendor to manage.

Why Teams Price with Hoonify
No markup mystery
Cloud AI prices are on the catalog, and that's what you pay. Sovereign is one annual subscription, agreed up front.
OpenAI-compatible
Point your existing SDK at Hoonify and keep your code, in our cloud or on your hardware. Switching cost: one base URL.
Your data stays yours
Nothing is kept or used beyond your work in our cloud. On-prem or air-gapped, it never leaves your network.
Fully managed
National lab engineers deploy, run, and support it, in our cloud or on your hardware. No extra ops team to hire.
Team accounts
Bring your whole team under one Hoonify account, in our cloud or on your hardware. One invoice, full visibility, no per-seat fees, and the access each person needs.
Every engineer gets their own API key under one organization. Usage rolls up to one account, whether it runs in our cloud or yours.
See usage by member, key, project, and model. In the cloud, finance sees what each workload costs. On-prem, you see who's using your capacity.
Admins manage billing, budgets, keys, and access. Developers just build.
Add unlimited team members free. You pay for usage or your subscription, not headcount.

Trust
Your data is yours
Never used for training.
Transparent marketplace
Prices come from live operator capacity
Billions of tokens served this week
Across the network
Questions, Answered
How am I billed?
In Hoonify Cloud, AI is credit-based: buy credits in advance, spend them at catalog rates, and top up when you need to. You can never owe more than you've loaded. HPC in our cloud is billed by usage. On your own hardware, Sovereign AI and Sovereign HPC are one annual subscription, priced per GPU or CPU node and invoiced. Talk to our team to scope your deployment
Is there a free tier?
Every new Cloud account gets $10 in credits for AI. Evaluating Sovereign? Start with those credits to test your workload before we size your deployment.
What counts as a cached token?
In Cloud, it's input that repeats context we've already processed, like a long system prompt. It's billed at a lower rate, with nothing to configure. On Sovereign, there's no per-token billing, so caching only makes things faster.
Do prices change?
Cloud prices can change, and the catalog always shows the current rate. Your Sovereign subscription price is fixed for the length of your contract.
Can I run models on my own hardware instead?
Yes. That's Sovereign. Our engineers deploy and manage Sovereign AI, Sovereign HPC, or both on-prem, in your private cloud, or fully air-gapped.
Is there volume or enterprise pricing?
Yes. Sovereign pricing drops as you deploy more GPUs or CPU nodes, and multi-year agreements save more. High-volume Cloud customers can also talk to us about committed-use rates. Sovereign pricing isn't published because it depends on your deployment, so we'll put together a quote for you.
Do you train on my data?
No. Your data is never used to train models. In our cloud, AI runs with zero data retention by default. On Sovereign, your data never leaves your network, so we never see it.
Can I start in the cloud and move on-prem later?
Yes, for AI and HPC, and most teams do. Prove the workload in Hoonify Cloud, then move it to your hardware. Your code and workflows stay the same.
One marketplace. Every operator running Hoonify. Sign up in minutes and send your first request today.