Per-token rates and TEE GPU deployments, priced in the open.
No middlemen. No markup tiers. Pay for the tokens you use, or deploy your own chute on a confidential GPU and pay by the second.
Model pricing
Pay per token. No subscription, no minimum, no markup. All featured models run on confidential TEE compute.
Estimate your cost
Pick a workload. Prices update below.
Pay from a Bittensor wallet in TAO
Looking for something else? Browse the full model catalog in the app.
Browse all modelsDeploy your own private chute
Run your own dedicated AI workload on verified self-serve confidential GPU capacity. Your code, your weights, your data, isolated end-to-end. Deployed in minutes from the chutes CLI. Use this when you need a private model, a custom fine-tune, or a dedicated instance for production traffic.
How it works
- Build your image, define your chute
Use the CLI to build a container, declare your
NodeSelectorwith the GPU class, VRAM, and count your workload needs, then expose your endpoints via cords. - Deploy with one command
chutes deployregisters your chute, pays the one-time fee, and brings it online as private by default. - Pay only while it runs
Billed by the second at the GPU's hourly rate. Idle instances shut down automatically, so no GPU-seconds are wasted.
# In your chute definition:
# choose the self-serve private GPU class
# tee=True, node_selector=NodeSelector(gpu_count=1, include=["pro_6000"])
chutes build my_chute:chute --wait
chutes deploy my_chute:chute --accept-feePrivate Chutes are billed at the GPU's hourly rate for however long the instance runs, plus a one-time deployment fee equal to 3× the hourly rate at the time of deployment. No subscription required.
GPUs shown here must have live pricing and TEE measurement support. The deployment fee is paid once when you run chutes deploy; after that, you pay only the per-second hourly rate while the instance is up. Actual placement follows your NodeSelector and available capacity.
Frequently Asked Questions
Have questions about our pricing? We've got answers.
How does billing work?
Billing is pay-as-you-go: top up your account balance and pay only for what you use (per-token rates for hosted models, per-second GPU time for Private Chutes), with no subscription and no minimum commitment.
How much do specific models cost?
Per-model token rates are listed in the Model pricing section above. Costs are charged per million tokens, separately for input and output, and pulled live from our pricing API, so what you see is what you pay.
What is a Private Chute and when would I want one?
A Private Chute is your own dedicated AI workload running on a confidential (TEE) GPU. Use it when you need to run custom code, a private fine-tune, or a model that is not in our public catalog, and you want full isolation and dedicated capacity.
How is the deployment fee for a Private Chute calculated?
Private Chutes are charged at the underlying GPU's hourly rate while running, plus a one-time deployment fee equal to 3× that hourly rate at the time of deployment. Pricing is per-second after that, so you only pay for actual compute time.
What happens to my existing monthly plan?
Monthly plans are no longer sold on chutes.ai. If you already have one, it stays active until the end of your current billing period and then does not renew. After that, your usage is billed pay-as-you-go from your account balance.
Do you offer volume discounts?
Yes, we offer custom pricing for high-volume users and enterprise customers. Contact our sales team for a personalized quote.
Is there a free trial?
We do not offer a free tier at this time. Top up your account in the app to get started. Pay-as-you-go has no minimum commitment.
What payment methods do you accept?
We accept 25+ payment methods via Stripe including crypto, cards, and bank transfers. We also accept payments in TAO and Bittensor alpha tokens, and for enterprise customers, we can arrange custom payment terms including invoicing.
Still have questions?
Contact Our Sales Team