AD
Replicate logo

Replicate

AcquiredFreemium🇺🇸Near-Breakeven

Run any AI model with a single line of code

84

Overall score

38

Heat score

Pricing

Free Trial$0
Pay-as-you-goUsage-based
EnterpriseCustom

Technical Specs

Inputs

Text Prompt, Image, Audio File, Video File, URL, Code, Training Data, Model Weights

Outputs

Generated Image, Generated Video, Generated Audio, Model Predictions, API Response, Fine-Tuned Model, Code Output, Text Output

AI Type

Agentic AI

Model Architecture

Custom/Proprietary

Daily Prompts

N/A

Context Length

N/A

Output Quality

Accuracy

84%

Content

88%

Reasoning

76%

Company Profile

Company

Replicate, Inc.

Founded

2019

HQ

San Francisco, California, USA

Employees

37

Total Raised / Total Funding

$57.8M

Revenue

$5.3M

Valuation

$350M

ARR

$4.8M

CEO

Ben Firshman

Overview

Estimated Paid Users

30K

Current estimate

Total Earnings Till Date

$8.5M

+3.30% from last month

Market Share

2.1%

Current share

Average Session

18

Per active user

Hallucination Rate

18%

Model quality signal

Growth Rate

+3.90%

Monthly active users

Burn Rate

$6.2M

Total expenses / years active

Paid User Gain

+12.50%

Monthly paid user trend

Profit Analysis

$15M

Total Loss

$4.6M

Total Profit

$24M

Performance Metrics

Accuracy

84%

Context

88%

Reasoning

76%

Safety

80%

Benchmarks

No benchmark scores available.

Replicate Models

Replicate Platform v1

Type: Other

Description: Initial version of the Replicate platform for hosting and running open-source ML models via REST API using Cog containers.

Architecture: Custom/Proprietary

Replicate Platform v2 (Generative AI Era)

Type: Multimodal

Description: Rebuilt infrastructure supporting Stable Diffusion and generative AI models at scale. Introduced GPU autoscaling and the public model catalog.

Architecture: Custom/Proprietary

Replicate Platform v3 (Cloudflare Integration)

Type: Multimodal

Description: Current version integrated into Cloudflare's global network. Features 50,000+ models, fine-tuning, deployments, and Workers AI integration.

Architecture: Custom/Proprietary

Funding Rounds & Investors

Total Funding

$57.8M

Rounds

3

Series B

$40M

Jun 2023

Led by a16z; post-money valuation $350M; 2M+ developers on platform; source: Replicate blog and FinSMEs 2023-12-06

Series A

$12.5M

Feb 2023

Part of $17.8M total raised; led by Sequoia; announced via TechCrunch 2023-02-21

Seed

$5.3M

Jan 2021

YC W20 batch seed funding; source: TechCrunch 2023-02-21 and The SaaS News

Founders/Team

BF

Ben Firshman

Co-Founder & CEO

AJ

Andreas Jansson

Co-Founder

Direct competitors

No direct competitors available.

Change Log / Major Updates

2024 · Dec 2

Migrated all A40 GPU workloads to faster NVIDIA L40S GPUs, delivering ~40% median speed improvement. H100 GPUs and multi-GPU configurations (2x, 4x, 8x A100/L40S) also made generally available.

2025 · Nov 17

Cloudflare announced acquisition of Replicate with 50,000+ models to be made available to Cloudflare Workers AI users. Replicate continues as a distinct brand.

2025 · Dec 1

Acquisition officially closed. Replicate team joins Cloudflare. API remains unchanged; integration with Cloudflare Workers, R2, and Durable Objects roadmap begins.

Compliance, Integrations & Support

Industry: Not specified

Compliances: Not specified

Integrations: Python SDK, JavaScript/Node.js SDK, Cloudflare Workers, Cloudflare Workers AI, Cog, Hugging Face, Stable Diffusion, Llama 3, Mistral, FLUX, Whisper, Claude, GPT-4, Vercel, Next.js, AWS, REST API, Google Colab, Zapier, GitHub

Support:email, help center, community forum, enterprise support

Target audience: Software Developers, AI Engineers, ML Researchers, Startups, Indie Hackers, Product Teams, Data Scientists, Creative Technologists

Supported languages: English

Replicate Acquisitions

CI

Cloudflare, Inc.

November 17, 2025

N/A

AD

Reviews & Rating

0 reviews

No reviews yet

Be the first to share how Replicate performs for your workflow.

0.0

Accuracy

0.0

Ease of Use

0.0

Output Quality

0.0

Security

0.0

More About Replicate

In the summer of 2022, Replicate co-founder Ben Firshman woke up sick with COVID and found something stranger on his laptop: a massive spike in traffic from developers worldwide rushing to the platform to run Stable Diffusion, the just-released open-source image model that had electrified the internet. Firshman and co-founder Andreas Jansson had launched Replicate the previous year, quietly building infrastructure for a world that wasn't quite ready yet—until suddenly it was.

What Replicate Does

Replicate is a serverless GPU cloud platform that lets any developer run over 50,000 open-source machine learning models—from Stable Diffusion to Llama-3 to Whisper—with a single line of Python or JavaScript. Instead of wrestling with CUDA drivers, dependency hell, and GPU provisioning, developers make one API call and Replicate handles provisioning, execution, scaling, and teardown. Billing runs by the second, only when a model is actually running. The company's open-source tool, Cog, defines a universal container format for packaging ML models, which became a de facto standard adopted across the industry.

  • Model catalog: 50,000+ public and community models, including image, video, audio, and language models
  • Custom deployment: Teams can deploy and scale their own fine-tuned models with no infrastructure management
  • Fine-tuning: Train models on custom data directly through the API
  • Deployments: Always-on dedicated endpoints with configurable autoscaling for production workloads

From Y Combinator to Cloudflare

The company went through Y Combinator's Winter 2020 batch—graduating right into a pandemic with no customers and no demo. Through years of quiet infrastructure building for niche ML communities, Replicate found its product-market fit when generative AI exploded in late 2022. It raised a $17.8M Series A in early 2023, then a $40M Series B led by Andreessen Horowitz in June 2023, reaching a $350M valuation with over 2 million developers and 30,000 paying customers. Clients including BuzzFeed, Unsplash, Character.ai, and Magnific built production applications on the platform. On November 17, 2025, Cloudflare announced it would acquire Replicate, and on December 1, 2025, the deal officially closed—with Replicate continuing as a distinct brand integrated into the Cloudflare Developer Platform.

Replicate FAQ's

What is Replicate?

Replicate is a serverless GPU cloud platform that lets developers run, fine-tune, and deploy machine learning models via a simple API. It hosts over 50,000 open-source models and was acquired by Cloudflare in December 2025.

How does Replicate pricing work?

Replicate charges by the second based on the GPU hardware used. Prices range from $0.000225/sec for T4 GPUs to $0.0032/sec for A100 (80GB) GPUs. You only pay while your model is actively running a prediction.

What is Cog?

Cog is an open-source tool created by Replicate that packages machine learning models into standard, reproducible containers. It defines a universal format for model inputs, outputs, and dependencies, making it easy to run models anywhere.

Can I deploy my own custom model on Replicate?

Yes. Using Cog, you can package your custom model and push it to Replicate. It will run on Replicate's GPU infrastructure, auto-scale with demand, and be accessible via a REST API.

What happened after Cloudflare acquired Replicate?

Replicate continues to operate as a distinct brand under Cloudflare. The API and all existing models keep working. Users will benefit from Cloudflare's global network for faster inference, and Replicate's model catalog is being integrated into Cloudflare Workers AI.