Replicate

What it does

Replicate lets developers run AI models through a cloud API. You send a request with model-specific input, such as a text prompt, an image, or a set of training images, and get back generated media or text. Hosted models cover image, video, speech and music generation, image restoration, image captioning and large language models.

Most people call a published model from Node.js, Python or plain HTTP with an API token, using a single run call. Beyond running existing models, you can fine-tune image models on your own images to produce a new model, or package custom code with Cog, Replicate's open-source packaging tool, and deploy it as an API. Replicate says it handles scaling, including scaling down to zero when idle, and offers logging and metrics. The catalogue mixes official models with community contributions.

It suits developers and product teams adding AI features without managing GPUs. You need an account and an API token, and the tool is used through code rather than a polished consumer app. Costs are usage-based and vary by model and hardware. Private custom models on dedicated hardware bill for setup and idle time as well as active time, which matters for low-traffic deployments. Replicate has joined Cloudflare, so check how that affects your plans.

What it can do

• Run hosted models for image, video, speech, music and text generation via API
• Fine-tune image models on your own training data
• Deploy custom models packaged with the open-source Cog tool
• Automatic scaling, including scale to zero when idle
• Logging and metrics for monitoring deployed models
• Client libraries for Node.js and Python plus an HTTP API

Example workflows

Generate an image from a prompt

Input: API token and a text prompt. Steps: install the Python or Node client, choose a published image model such as black-forest-labs/flux-dev, call run with prompt and options like aspect ratio. Output: a generated image file URL.

Fine-tune an image model on custom images

Input: a zip of training images and a trigger word. Steps: create a training with a trainer model and a destination, wait for completion, then run the new model. Output: a model that generates images of your subject or style.

Before you choose

• Billing is per second of hardware time or per input/output depending on the model, so check each model page for cost estimates
• Private custom models on dedicated hardware bill for setup and idle time, except fast-booting fine-tunes
• Replicate is now part of Cloudflare, so future product direction may change

Listed under

• Coding
• Sandboxes and infrastructure
• Developer tools
• Fine-tuning
• GPU
• Image generation
• Inference API
• MLOps

Get started

API token needed to call models

1. Create an account at replicate.com/signin
2. Generate an API token and store it as REPLICATE_API_TOKEN
3. Install the Node or Python client and run a model, or follow the docs to push a custom model with Cog

Public models bill by run time or by input/output depending on the model. Private models on dedicated hardware bill for setup, idle and active time. See the hardware table for per-second rates. Enterprise volume discounts exist.

Source: replicate.com/pricing, checked 29 Oct 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Replicate - Run AI with an API ↗replicate.com
• Documentation – Replicate ↗replicate.com
• Pricing – Replicate ↗replicate.com
• Why Replicate is joining Cloudflare | Cloudflare Blog ↗blog.cloudflare.com

Company website

replicate.com: product, pricing and policy pages

Hacker News

Launch posts and discussion

YC directory mirror

Batch and company name

Human review

Checked by a person

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.