OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Infrastructure & Serving

Openpipe

OpenPipe is an open-source fine-tuning and model-hosting platform that uses past LLM requests to develop smaller models for specific tasks. It documents querying request logs, importing OpenAI-compatible JSONL datasets, evaluating model outputs and hosting fine-tuned models or downloading their weights. Python and TypeScript SDKs, along with an OpenAI-compatible chat-completions endpoint, support integration with exi

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Openpipe — OpenViST project profile in AI Infrastructure & Serving.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Consider OpenPipe when you have request history or compatible training data and want to evaluate whether a smaller fine-tuned model can handle a specific task. Its documented local setup is relevant when you want to develop against a self-hosted instance.

Who it is for

  • Developers working with OpenAI-style chat-completion workflows who want to evaluate task-specific fine-tuned models.
  • Teams with LLM request history or OpenAI-compatible JSONL datasets to use in model evaluation and fine-tuning.
  • Developers who need to run the platform locally and can configure PostgreSQL, Node.js, pnpm and GitHub OAuth.

Documented uses

  • Query logged LLM requests using filters.
  • Import OpenAI-compatible JSONL data and prepare it by pruning duplicated text such as system prompts.
  • Fine-tune supported models using task-specific request data.
  • Compare model outputs against past requests or a base model.
  • Host fine-tuned models on-platform or download their weights.
Before you choose

Requirements & limits

  • The repository documents Python and TypeScript SDKs and an OpenAI-compatible chat-completions endpoint.
  • Documented data workflows include request-log filtering, OpenAI-compatible JSONL import and pruning duplicated text.
  • The README lists GPT-3.5, Mistral and Llama models and says fine-tuned weights can be hosted on-platform or downloaded.
  • Local setup requires PostgreSQL, Node.js 20, pnpm, an OpenAI API key and GitHub OAuth configuration.
  • The README says development on the open-source version is temporarily stopped while proprietary third-party code is integrated.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up OpenPipe for local development

Prepare and start a local OpenPipe instance using the repository's documented setup sequence.

Before you start

  • Node.js 20
  • pnpm
  • PostgreSQL
  • An OpenAI API key
  • A GitHub OAuth app with a callback URL for the local instance

Steps

  1. Clone the repository and install its dependencies with pnpm.

  2. Copy app/.env.example to app/.env and enter the OpenAI API key.

  3. Set DATABASE_URL to the PostgreSQL instance and run pnpm prisma migrate dev in the app directory.

  4. Set GITHUB_CLIENT_ID and GITHUB_CLIENT_SECRET using the GitHub OAuth app, then run pnpm dev in the app directory.

Expected outcome

The README directs users to open the local application at http://localhost:3000.

Guide limitations

  • The README says development on the open-source version is temporarily stopped while proprietary third-party code is integrated.
  • The documented local setup requires GitHub OAuth configuration.
Documentation 1 ↗
Source-based practical guide

Send and log a chat completion to a local instance

Use the documented Python example to submit a chat completion to a local OpenPipe instance and log the request.

Before you start

  • A running local OpenPipe instance
  • The OpenPipe Python package
  • An OpenAI API key and OpenPipe API key

Steps

  1. Create an OpenAI client configured with the OpenAI key, OpenPipe key and local base URL http://localhost:3000/api/v1.

  2. Call client.chat.completions.create with a model and messages.

  3. Set openpipe.log_request to True and provide a prompt_id tag in the request's OpenPipe options.

Expected outcome

The example submits a chat completion and requests that the local instance log it with the supplied prompt_id tag.

Guide limitations

  • The README provides an example request, not a complete guide to fine-tuning or serving a resulting model.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
OpenPipe/OpenPipe · Official homepage title · retained in its source language
How we review sourcesSuggest a correction