Replit AI Integrations

Documentation Index

Fetch the complete documentation index at: https://docs.replit.com/llms.txt

Use this file to discover all available pages before exploring further.

Replit AI Integrations makes it easy to build AI-powered applications. Get started with models from OpenAI, Anthropic, Google, and OpenRouter immediately—no developer account setup, no API key management. Replit handles everything for you.

​

What is Replit AI Integrations?

Replit AI Integrations provides managed access to AI models from leading providers. Build chat apps, content generators, and AI-powered tools without leaving Replit. No need to create external developer accounts, find your API credentials, or manage separate billing. When you use Replit AI Integrations, Replit manages the provider credentials and bills you at the public API price. Your usage is billed to your Replit credits and appears on your usage page, broken down by Existing Replit App.

​

Features

Replit AI Integrations streamlines AI development with built-in access and transparent billing:

• Zero setup required: No developer accounts or API keys needed to get started
• Managed credentials: Replit handles provider authentication and credential management
• Transparent billing: Pay public API prices, billed to your Replit credits
• Usage tracking: View your usage per Existing Replit App on your usage page
• Cross-stack support: Works across all programming languages and frameworks
• Optional BYOK: Bring your own API key if you prefer to use your own credentials

​

Getting started

​

Using Replit-managed credentials

When you mention AI functionality or specific providers (OpenAI, Anthropic, Google, or OpenRouter) in your prompt, Agent defaults to using Replit AI Integrations: You can tell which integration is being used by looking for “OpenAI (Replit managed)”, “Anthropic (Replit managed)”, “Google (Replit managed)”, or “OpenRouter (Replit managed)” in your initial prompt response. This indicates Agent will use Replit’s credentials instead of requiring your own API key. For initial builds: When you submit an initial prompt to create a new app with AI functionality, Replit AI Integrations is matched automatically. Agent will start building with Replit’s managed credentials right away. For existing Replit Apps: When Agent detects you want to add AI functionality to an existing Replit App, Agent will explicitly tell you in the Project Editor that it’s asking to use Replit AI Integrations instead of your own API key. You’ll see a confirmation prompt: Click “Approve” to continue with Replit-managed credentials, or “Dismiss” if you prefer to use your own API key. Your usage is automatically tracked and billed to your Replit account at the public API price.

​

Bringing your own API key

If you prefer to use your own credentials from OpenAI, Anthropic, Google, or other providers instead of Replit AI Integrations, you have a few options: For initial builds: Include in your prompt that you want to use your own API key: You’ll see the provider name (without “Replit managed”) in the response, indicating Agent will request your API key. For existing Replit Apps: When Agent tells you it wants to use Replit AI Integrations and shows the confirmation prompt, click “Dismiss” instead of “Approve.” In both cases, Agent will build through the first checkpoint before requesting your API key. This allows Agent to set up your project structure first, then pause to collect your credentials. When you use your own API key, you’re billed directly by the provider rather than through Replit credits.

​

Viewing your usage

All Replit AI Integrations usage appears on your usage page alongside your other Replit usage. Usage is broken down by Existing Replit App so you can see exactly where your AI spend is going.

​

Supported providers

Replit AI Integrations currently supports:

• OpenAI: Multimodal models for text generation, reasoning, and audio. Supports chat completions, responses, image generation, audio transcription (gpt-4o-transcribe, gpt-4o-transcribe-mini), and audio output (gpt-4o-audio, gpt-4o-audio-mini).
• Anthropic: Multimodal models for reasoning, writing, and coding. Supports the Messages API and interactions with web search.
• Google (Gemini): Multimodal models for text generation and reasoning. Supports text and image generation.
• OpenRouter: Access to models from providers such as Microsoft (Phi series), Meta (Llama series), Mistral, Qwen, DeepSeek, Nvidia, Amazon, and more.

​

OpenRouter privacy and availability

When you use OpenRouter through Replit AI Integrations, your requests are sent with specific privacy settings to protect your data. Understanding these settings is important because they may affect which models are available to you.

​

Privacy settings for self-serve users

For individual and team users, Replit configures OpenRouter requests with the following privacy defaults:

• Paid endpoints training disabled: Your data will not be used for training by paid model providers
• Free endpoints training enabled: Free model providers may train on your prompts and completions
• Free endpoints publishing enabled: Free providers may publish your prompts and completions to public datasets
• Input/output logging disabled: Your requests and responses are not logged by OpenRouter for the 1% billing discount
• Analytics cookies disabled: Analytics cookies are not used

These settings prioritize your privacy while enabling access to the widest range of models. However, some models may not be available if they require different privacy settings than those configured above.

​

Privacy settings for Enterprise

For Enterprise organizations, Replit enforces stricter privacy controls:

• Zero Data Retention (ZDR) endpoints only: Only models with a Zero Data Retention policy are available
• All privacy settings listed above for self-serve users also apply
• This ensures that Enterprise requests are only routed to endpoints that do not retain any data

Because Enterprise enforces ZDR endpoints only, the selection of available models will be more limited compared to self-serve users. This trade-off ensures the highest level of data privacy for enterprise workloads.

​

Teams and Enterprise

For Teams and Enterprise organizations, Replit AI Integrations access is disabled by default. Organization admins can control access from the organization settings page:

1. Navigate to your organization settings
2. Find the Replit AI Integrations section
3. Toggle access on or off for your organization

When enabled, organization members can use Replit AI Integrations in their apps. All usage is tracked and billed to the organization account. When disabled, members will need to provide their own API keys to use AI model providers.

​

Use cases

Build powerful AI applications across any stack:

• Content generation: Summarize articles, generate descriptions, create content
• Analysis and insights: Extract key information, analyze sentiment, classify text
• Conversational interfaces: Build chatbots and AI assistants
• Text transformation: Translate, rewrite, or format content
• Code assistance: Generate code snippets, explain algorithms, review code

​

FAQ