Using GRPO to beat o3-mini at Clue

GRPO beats o3-mini at Clue →

OpenPipe for Research Agents

Power the most advanced research agents with innovative fine-tuning techniques on OpenPipe.

State of the art

Increase agent success rates beyond what prompted models alone can achieve.

Advanced tool calls

Teach your agent how to use the tools at its disposal to solve complicated real-world tasks.

Domain expertise

Adapt your research agents to niche domains or usage patterns and watch them excel.

Built for Enterprise Research Agents

Your models, your way

___

Ditch reliance on third-party models. Use OpenPipe to easily create custom models for deep research applications for increased speed, reliability, and enhanced quality. Create unlimited models for each step of your pipeline and deploy on-prem or via OpenPipe's servers.

Dedicated engineering support

___

OpenPipe's enterprise solution architects will guide you every step of the way. Get unparalleled support with custom SLAs and feature prioritization, with flexibility to leverage OpenPipe's internal engineering team for custom builds.

Save more as you scale

___

In addition to the significant cost savings on inference you get from customizing your own models, OpenPipe offers additional discounts as you scale. For more information on customized pricing plans, please schedule an enterprise demo with our team below.

Book an enterprise demo

Sign up to our newsletter

Stay updated with our latest product releases!

About OpenPipe

OpenPipe is the easiest way to train and deploy your own fine-tuned models. It only takes a few minutes to get started and can save you 25x relative to OpenAI with higher quality.

Sign up to our newsletter

Stay updated with our latest product releases!

About OpenPipe

OpenPipe is the easiest way to train and deploy your own fine-tuned models. It only takes a few minutes to get started and can save you 25x relative to OpenAI with higher quality.