AI Providers

Configure AI providers for enhanced transcription

Supported Providers

Pipit supports AI providers for post-processing your transcriptions:

OpenRouter

Access to Google Gemini, Claude, and many other models through a unified API.

Default model: google/gemini-2.5-flash-lite

Custom Endpoint

Any OpenAI-compatible API, including local models on your machine.

Recommended for: Privacy-focused users, offline use, cost savings

Why Use Local Models?

Running AI models locally offers three key advantages: Privacy, Cost, and Offline Use. Your transcriptions never leave your machine, and you don’t pay per-token fees.

Local Model Guide

Learn how to set up Ollama, LM Studio, and other local inference servers for Pipit.

OpenRouter

For cloud-based AI without managing infrastructure, OpenRouter provides a unified API to access dozens of models including Google Gemini, Anthropic Claude, and open-source models.

1. Get API Key

   1. Go to openrouter.ai and sign in
   2. Create an API key from your account settings
   3. Copy the key (you will not see it again)
2. Configure Pipit

   1. Open Pipit Settings → AI Processing
   2. Select OpenRouter as the provider
   3. Paste your API key
   4. Choose a model (google/gemini-2.5-flash-lite offers good speed/cost balance)

Custom Endpoints (Generic)

Any service that provides an OpenAI-compatible API works with Pipit. This includes local inference servers as well as specialist cloud providers.

Custom endpoints require a valid base URL. API keys are optional and can be left blank for local servers.

Configuration

To use a custom endpoint, select Custom Endpoint in Pipit Settings:

Setting	Description
Endpoint URL	The base URL of the API (e.g., https://api.together.xyz/v1 or http://localhost:11434/v1)
API Key	Your secret API key (leave blank for local servers)
Model Name	The specific model identifier to use

Common Compatible Services

Service	URL Format	Notes
Ollama	http://localhost:11434/v1	See Local Model Guide
LM Studio	http://localhost:1234/v1	See Local Model Guide
Together AI	https://api.together.xyz/v1	High-performance open-source models
Fireworks	https://api.fireworks.ai/v1	Fast, reliable serverless inference
LocalAI	http://localhost:8080/v1	Docker-based self-hosted AI

Timeouts

• OpenRouter: 3 seconds for fast fallback to raw transcription.
• Custom endpoints: 15 seconds to accommodate slower local inference.

Troubleshooting

For issues specific to running models on your own machine, please refer to the Local Model Troubleshooting section.