# Oumi

> Oumi (Open Universal Machine Intelligence) is a Kirkland/Bellevue, Washington-based public benefit corporation building a fully open-source, end-to-end platform for foundation models. Founded in 2024 by ex-Google, Apple, Meta and Microsoft engineers, its open software stack lets researchers and enterprises prepare data, pretrain, fine-tune, evaluate and deploy LLMs and vision-language models - from 10M to 405B parameters - anywhere from a laptop to a cloud cluster. Backed by $10M in seed funding and a coalition of 13 top research universities, Oumi pitches itself as the 'Linux of AI': a neutral, collaborative, unconditionally open alternative to closed frontier labs.

- **Founded:** 2024
- **Headquarters:** Bellevue, Washington, United States
- **Founders:** Emmanouil (Manos) Koukoumidis (CEO & Co-founder), Oussama Elachqar (Co-founder, Engineering), Matthew Persons (Co-founder, Engineering), William Zeng (Co-founder, Engineering), Jeremiah Greer (Co-founder, Research), Kostas Aisopos (Co-founder, Research)
- **Team size:** ~40 employees
- **Products:** Oumi Open-Source Platform, Recipes & Model Zoo, Evaluation & Judges, Deployment & Cloud, Enterprise Platform
- **Notable:** Launched what it calls the world's first 'unconditionally open' AI platform out of stealth in January 2025., Raised $10M seed led by Venrock and Obvious Ventures., Built a founding coalition of researchers from 13 top universities.

## Products & services

- **Oumi Open-Source Platform** — An end-to-end, fully open-source framework for building foundation models covering data prep, pretraining, fine-tuning (SFT, LoRA, QLoRA, GRPO), evaluation, and deployment. Supports text and multimodal models from 10M to 405B parameters.
- **Recipes & Model Zoo** — 200+ ready-to-use recipes and 100+ supported models (Llama, DeepSeek, Qwen, Phi and more) so teams can get started in minutes with zero boilerplate - no custom training loops or data pipelines required.
- **Evaluation & Judges** — Comprehensive model benchmarking across standard evals, plus data synthesis and curation using LLM judges to build and clean training datasets.
- **Deployment & Cloud** — Deploy with popular inference engines (vLLM, SGLang) and run anywhere - laptops, clusters, or clouds including AWS, Azure, GCP and Lambda.
- **Enterprise Platform** — A managed, enterprise-grade offering with role-based access, model management, auto-updates and cost reduction for organizations building and maintaining custom AI in-house.

## Achievements

- Launched what it calls the world's first 'unconditionally open' AI platform out of stealth in January 2025.
- Raised $10M seed led by Venrock and Obvious Ventures.
- Built a founding coalition of researchers from 13 top universities.
- Assembled a team of engineers who worked on Google Gemini, Apple Intelligence, and Google Cloud GenAI.
- Shipped 200+ recipes and support for 100+ models in an open framework spanning 10M to 405B parameters.

## Latest updates

- **2025-01** — Came out of stealth with $10M seed funding and unveiled its unconditionally open AI platform.
- **2025-01** — Announced founding collaboration with researchers from 13 leading research universities.

## Links

- Website: https://oumi.ai
- LinkedIn: https://www.linkedin.com/company/oumi-ai
- Twitter/X: https://twitter.com/Oumi_PBC
- GitHub: https://github.com/oumi-ai/oumi
- YouTube: https://www.youtube.com/@oumi_ai

---

Profile page: https://yespress.io/oumi
Published by YesPress — https://yespress.io
Last updated: 2026-07-03