# OpenPipe Projects at AI Tinkerers

> Canonical HTML: https://aitinkerers.org/technologies/openpipe
> Markdown URL: https://aitinkerers.org/technologies/openpipe.md
> Technology record last updated: 2026-03-05T10:05:07Z
> Generated: 2026-09-22T04:37:47Z

Turn expensive, general-purpose LLM prompts into cheap, high-performance fine-tuned models.

OpenPipe is the developer platform and SDK for production-ready AI, abstracting the complexity of large language model fine-tuning. The platform captures existing prompt-completion pairs from your current provider (e.g., OpenAI) and uses that production data to train a new, specialized model. This process delivers immediate, measurable results: users routinely see up to 8x lower inference cost than GPT-4-class APIs and often achieve higher accuracy on specific tasks like data extraction or classification. OpenPipe supports a range of base models (Llama, Mistral, Qwen) and includes advanced tools like the Agent Reinforcement Trainer (ART) for continuous, data-driven optimization of AI agents.

- Official technology site: https://openpipe.ai
- Public AI Tinkerers demos and talks: 4
- Result page: 1 of 1

## Recent Public Talks and Demos

### [Training Practical Agents with Reinforcement Learning](https://seattle.aitinkerers.org/talks/rsvp_IHaOyIAK0WU)

Covered in the talk: 1) how to get started with reinforcement learning, 2) how we trained an email agent to beat o3, 3) how to train a SOTA summary agent during this hackathon

- Event context: CrewAI x AI Tinkerers - Practical Agents Hackathon - Seattle — 2025-05-17 — Seattle
- Public talk page: https://seattle.aitinkerers.org/talks/rsvp_IHaOyIAK0WU

### [LABEL: LLM Assistance for Better Evaluation Labels](https://seattle.aitinkerers.org/talks/rsvp_kadYvSmadkQ)

A demo workflow and UX for labeling data, using it to evaluate LLM-evaluators, and then aligning the LLM-evaluator to human judgments (and perhaps optimizing the evaluator!) P.S., Kyle Corbitt of OpenPipe will be demoing something similar, and I hope to have a faceoff with them by going before them so the audience can decide the pros and cons of each.

- Event context: AI Tinkerers - Seattle - October 2024 Meetup — 2024-10-25 — Seattle
- Public talk page: https://seattle.aitinkerers.org/talks/rsvp_kadYvSmadkQ

### [Cross-training models on multiple providers](https://seattle.aitinkerers.org/talks/rsvp_Z0XdL4RVkyw)

OpenPipe is releasing an open-source library to streamline the process of fine-tuning an LLM on any provider with minimal effort. We've personally benefited from the open-source community in numerous ways, and we'd like to show developers how easy it is to use the latest tools to save massive amounts on compute costs.

- Event context: AI Tinkerers Seattle - April 2024 Meetup — 2024-04-26 — Seattle
- Public talk page: https://seattle.aitinkerers.org/talks/rsvp_Z0XdL4RVkyw

### [Fine-tune your own Llama 2 to replace GPT-3.5/4](https://seattle.aitinkerers.org/talks/rsvp_i9TfjPwwdKk)

I recently put together a set of Python code files that walk you through how to replace GPT-3.5 with your own fine-tuned Llama 2 model. The replacement model is 1/50th the cost and often more accurate. You can see the pretty active thread where I shared it with HN here: https://news.ycombinator.com/item?id=37484135 This isn't a pitch for our company or anything -- in fact the project only uses a small piece of OpenPipe functionality for convenience and is mostly standalone.

- Event context: AI Tinkerers Seattle - September Meetup — 2023-09-20 — Seattle
- Public talk page: https://seattle.aitinkerers.org/talks/rsvp_i9TfjPwwdKk

## Related Technologies

- [GPT-4](https://aitinkerers.org/technologies/gpt-4) ([Markdown](https://aitinkerers.org/technologies/gpt-4.md)) — 529 public demos
- [GPT-3](https://aitinkerers.org/technologies/gpt-3) ([Markdown](https://aitinkerers.org/technologies/gpt-3.md)) — 191 public demos
- [Llama-2](https://aitinkerers.org/technologies/llama-2) ([Markdown](https://aitinkerers.org/technologies/llama-2.md)) — 227 public demos
- [BERT](https://aitinkerers.org/technologies/bert) ([Markdown](https://aitinkerers.org/technologies/bert.md)) — 179 public demos
- [BLOOM](https://aitinkerers.org/technologies/bloom) ([Markdown](https://aitinkerers.org/technologies/bloom.md)) — 115 public demos
- [Data labeling](https://aitinkerers.org/technologies/data-labeling) ([Markdown](https://aitinkerers.org/technologies/data-labeling.md)) — 1 public demo
- [Fine-tuning](https://aitinkerers.org/technologies/fine-tuning) ([Markdown](https://aitinkerers.org/technologies/fine-tuning.md)) — 20 public demos
- [Human judgments](https://aitinkerers.org/technologies/human-judgments) ([Markdown](https://aitinkerers.org/technologies/human-judgments.md)) — 1 public demo
- [LangChain](https://aitinkerers.org/technologies/langchain) ([Markdown](https://aitinkerers.org/technologies/langchain.md)) — 445 public demos
- [Library](https://aitinkerers.org/technologies/library) ([Markdown](https://aitinkerers.org/technologies/library.md)) — 4 public demos
- [LLM Evaluators](https://aitinkerers.org/technologies/llm-evaluators) ([Markdown](https://aitinkerers.org/technologies/llm-evaluators.md)) — 1 public demo
- [OpenAI API](https://aitinkerers.org/technologies/openai-api) ([Markdown](https://aitinkerers.org/technologies/openai-api.md)) — 520 public demos
- [Open Source](https://aitinkerers.org/technologies/open-source) ([Markdown](https://aitinkerers.org/technologies/open-source.md)) — 12 public demos
- [PaLM 2](https://aitinkerers.org/technologies/palm-2) ([Markdown](https://aitinkerers.org/technologies/palm-2.md)) — 116 public demos
- [Python](https://aitinkerers.org/technologies/python) ([Markdown](https://aitinkerers.org/technologies/python.md)) — 662 public demos
- [PyTorch](https://aitinkerers.org/technologies/pytorch) ([Markdown](https://aitinkerers.org/technologies/pytorch.md)) — 273 public demos
- [RoBERTa](https://aitinkerers.org/technologies/roberta) ([Markdown](https://aitinkerers.org/technologies/roberta.md)) — 118 public demos