Home/Large language models/Replicate

Replicate pricing

Replicate is a hosted platform for running, fine-tuning and deploying machine learning models through an API.

Large language modelsPriced per minuteSelf-serve + sales

Visit ReplicatePrices checked Oct 29, 2026

How much does Replicate cost?

Usage price from $0/minute

Free plan: Not stated
Usage price: $0/minute
Enterprise: Contact sales

Read from replicate.com/pricing on Oct 29, 2026. Price changed? Tell usSee the line each price came from

In your favor

Notice before price changes. Reasonable prior notice of subscription fee changes

How you buy and get started

Self-serve + salesButtons on their site: Try for free, Get started for free, Talk to us

Sign up: Self-serve sign in, including with GitHub
Setup: Not stated
When sales steps in: Enterprise tier for volume discounts and SLAs via contact form
Support: Priority support for enterprise
Support: Support email

What Replicate does

Replicate lets developers run thousands of community and official models (image, video, speech, music and language) via a single API, fine-tune them, and deploy custom models packaged with Cog. Billing is pay-as-you-go by hardware time or by input/output, with an enterprise tier for volume discounts, SLAs and support.

Good for: Software developers and businesses that want to call or deploy AI models through an API without managing infrastructure.
Channels: API, Web playground, Node client, Python client, HTTP
Main product: Hosted API for running, fine-tuning and deploying AI models
Also includes: API access to thousands of community and official models, Fine-tuning of models with your own data, Custom model deployment with Cog, Automatic scaling, Playground for comparing models, MCP server and agent skills
Connects to: GitHub
Public API: Yes
MCP server: Yes

Public API accessRun AI with an API.
Long context windowNot stated
Fine-tuning supportRun and fine-tune models.
Tool and function callingNot stated
Image and audio inputNot stated
SOC 2 complianceNot stated
Zero data retention optionNot stated
VPC or on-prem deploymentNot stated

The same 8 checks appear on every tool in Large language models. Each tick quotes the vendor's own site.

Security and your data

Question	Their answer	Where it says so
Trains AI on your data?	Not stated
HIPAA (signs a BAA)	Not stated
SOC 2	Not stated
GDPR	Not stated
CCPA	Not stated
PCI	Not stated
ISO 27001	Not stated
Single sign-on	Not stated
Trust center	Not stated

Security answers are the vendor's own claims, quoted. We don't audit them.

Every plan

Plan	Monthly	Billed yearly	Included	Over the allowance	Where it says so
Hardware: CPU (Small)/sec	$0per minute	–	–	–	"$0.000025/sec"
Hardware: CPU/sec	$0per minute	–	–	–	"$0.000100/sec"
Hardware: Nvidia T4 GPU/sec	$0per minute	–	–	–	"$0.000225/sec"
Hardware: Nvidia L40S GPU/sec	$0per minute	–	–	–	"$0.000975/sec"
Hardware: Nvidia A100 (80GB) GPU/sec	$0per minute	–	–	–	"$0.001400/sec"
Hardware: Nvidia H100 GPU/sec	$0per minute	–	–	–	"$0.001525/sec"
Enterprise	Quote	–	–	–	"We've also got volume discounts for large amounts of spend."

Fine print

Billing increment: per second
Billed beyond usage: Not stated
Contract: Not stated
Refunds: Not stated
Auto renew: Subscriptions renew automatically for successive periods of the same duration unless cancelled
Price change notice: Reasonable prior notice of subscription fee changes