OpenPipe Analysis: $7M Raised
What is OpenPipe?
OpenPipe builds software for post-training large language models with supervised fine-tuning and reinforcement learning. Its ART open-source trainer and hosted platform help teams evaluate, fine-tune, and serve models while targeting lower latency and lower inference costs. The company says it is joining CoreWeave and positions its enterprise offering around on-prem deployment, compliance controls, and dedicated support.
Location
Seattle, United States
Employees
1-10
Founded
2023
Revenue
<$5M
Product Features & Capabilities
- Fine-tuning for production apps, Direct Preference Optimization (DPO) support, higher-quality and faster models, automatic data capture, two-click model training, automated deployment, LLM-as-judge evaluations, intuitive data and model management, flexible pricing plans, SOC 2 Type 2, HIPAA, GDPR compliance, ownership of fine-tuned model weights.
Pricing
-
Developer Plan:
- Per-Token pricing
- From $0.48 per 1M tokens for training
- $0.30 for input and $0.45 for output
- Includes autoscaling, metrics & analytics
- 50k training rows per dataset
- Up to 50 fine-tuned models
-
Business Plan:
- Enterprise pricing
- Includes everything in Developer Plan
- HIPAA, SOC 2, GDPR compliance
- Custom relabeling techniques
- Active learning
- 500k training rows per dataset
- Discounted token rates
- Unlimited fine-tuned models
How much OpenPipe raised
Seed - $6.7 million
March 25, 2024Lead Investor: Costanoa Ventures, Y Combinator
Who are the founders of OpenPipe
KC
Kyle Corbitt
Founder
DC
David Corbitt
Founder
Market Focus
OpenPipe primarily focuses on the artificial intelligence and machine learning industry, specifically in providing tools for training and deploying fine-tuned models.
Investors
- Costanoa Ventures
- Y Combinator
- Logan Kilpatrick (angel investor)
- Three other unnamed investors
Business Model
Charges for fine-tuning models and usage in production; flexible pricing based on tokens.
Competitors Overview
OpenPipe operates in a competitive landscape with several notable players in the AI model training and deployment market. The main competitors include:
-
Scale AI:
- Overview: Scale AI provides a platform for data labeling and model training, focusing on high-quality data for machine learning applications.
- Advantages: Scale AI is known for its extensive data annotation capabilities and robust infrastructure, which can handle large datasets efficiently. It offers a comprehensive suite of tools for data preparation, making it easier for companies to train their models effectively.
-
Together AI:
- Overview: Together AI offers an AI acceleration platform that allows users to run and fine-tune generative AI models with scalable infrastructure.
- Advantages: Together AI emphasizes ultra-fast inference capabilities and flexible deployment options. It claims to provide optimized performance at a lower cost compared to traditional closed-source providers, making it an attractive option for businesses looking to implement AI solutions quickly and cost-effectively.
Notable Differences:
- OpenPipe focuses on enabling users to fine-tune their models easily and cost-effectively, claiming significant savings compared to traditional models like GPT-4. In contrast, Scale AI is more centered on data preparation and annotation, while Together AI highlights its speed and scalability in deploying generative models.
See something that needs updating? Suggest edits to this profile.