Oumi · since 11. August 2026 (Compounding AI Factory Launch); Kernbibliothek erstveröffentlicht Januar 2025 · 3× · last seen Aug 15, 2026
Oumi is an enterprise AI platform (not hardware) that automates the entire lifecycle of specialized AI models — evaluation, data synthesis, training, deployment, and continuous retraining on production data. Founded in 2025 in Seattle/Bellevue by former Google Gemini, Apple, and Microsoft leaders, the company launched its "Compounding AI Factory" feature set on August 11, 2026, enabling one-click deployment to auto-scaling GPUs and turning production failures into retraining signals. The underlying core library is available as open-source under the Apache 2.0 license on GitHub, while the hosted platform offers usage-based/enterprise pricing with free starting credits.
Features
| License | Apache License 2.0 (core library, open source) |
| Platform | Cloud-based enterprise AI platform; runs on AWS, Azure, GCP, Lambda, etc.; models from 10M to 405B parameters |
| Price | Free starting credits: $50 (corporate) / $25 (personal); then usage-based; enterprise plans with dedicated support |
| Release Date | Compounding AI Factory launched August 11, 2026 |
| Memory | GPU resources auto-scale with traffic, scaling to zero when idle (no fixed memory spec given) |
| Availability | Generally available; sign-up with free credits via oumi.ai; CLI publicly available |
