Replicate: pricing, free plan, features & alternatives (2026)

What is Replicate?

Replicate is a cloud platform where you run machine learning models through an API instead of hosting them yourself. Its catalogue contains thousands of models contributed by the community, plus a range of proprietary models from companies such as Google, OpenAI and Black Forest Labs. You call a model from code (Node, Python or plain HTTP) by passing a model name and an input object, and you receive the output back.

The platform covers image generation and editing, speech, music, video generation, image restoration, image captioning and large language models. You can also deploy your own custom models using Cog, the open-source packaging tool, and fine-tune supported models. Public models are billed by run time, per output image, per second of video or per token, while private models run on dedicated hardware billed by the hour or second.

Key features

Run thousands of open-source and proprietary models through a single API
Client libraries for Node, Python and plain HTTP
Image generation and editing, including multi-image fusion and conversational edits
Video generation from text or images, with up to 30-second clips in some models
Fine-tuning and deployment of custom models with Cog
Dedicated GPU hardware for private models with automatic scaling

Who is it for

Developers integrating AI models into applications
Startups and product teams needing production-ready model APIs
Machine learning researchers and model publishers
Companies wanting to host custom models on dedicated hardware

How to get started with Replicate

Create an account on the Replicate website and generate an API key
Choose a model from the catalog, e.g. FLUX, and check its input parameters
Call the model with one line of code in Node, Python, or over HTTP

Choose it if

You are building an AI feature into an app and want models via API
You need image, video, speech, or music generation without maintaining your own infrastructure
You want to pay only for actual use of public models

Skip it if

You need a graphical interface in Polish, which the platform does not offer
You are looking for a fully free solution, since the site does not detail the free start limits

Pros and cons

Pros

Pay only for what you use, with per-second or per-output pricing for public models
Large catalogue of ready-to-use, production-ready model APIs

Cons

Private models bill for idle time as well as active time
Requires programming to use the API

Replicate pricing

Plan	Price	Includes
Public models (pay per use)	0.025 USD / one-time	Example: flux-dev, per output image; prices vary by model and are shown on each model's page
Private models - CPU (Small)	0.09 USD / one-time	Per hour, dedicated hardware, billed while online including idle time
Private models - Nvidia T4 GPU	0.81 USD / one-time	Per hour, dedicated hardware
Private models - Nvidia L40S GPU	3.51 USD / one-time	Per hour, dedicated hardware
Private models - Nvidia H100 GPU	5.49 USD / one-time	Per hour, dedicated hardware
Private models - Nvidia A100 (80GB) GPU	5.04 USD / one-time	Per hour, dedicated hardware

Prices copied from Replicate pricing, checked October 12, 2026. Currency as shown by the vendor; taxes may apply.

Replicate pricing details →

Is Replicate free?

There is a free trial. A 'Try for free' option is offered on the homepage; the text does not specify the free allowance.

Replicate alternatives

Tool	Price from	Free plan	Category
fal	0.05 USD	?	APIs & models
DeepAI	9.99 USD/mo	yes	APIs & models
Grok API	0.02 USD	?	APIs & models
Google AI Studio	0,75 USD	?	APIs & models
Z.AI Developer Platform	0.015 USD	trial	APIs & models
MiniMax	0.3 USD input / 1.2 USD output per M tokens	?	APIs & models