Map → L6 AI Runtime, Serving & Model Access → Predibase
Predibase
L6 · AI Runtime, Serving & Model Access
inference & servingstartupprivateseries A
Predibase is a developer platform designed for fine-tuning and serving open-source large language models on managed cloud compute. Leveraging open-source Ludwig and LoRAX technologies, it allows engineering teams to train task-specific adapter models on proprietary data. The system dynamically serves hundreds of fine-tuned adapter variants on single shared GPU instances.
Ecosystem functionLowers fine-tuning and deployment costs by serving multiple LoRA adapter variants on single shared GPU instances.
Business modelSaaS platform tier and runtime compute fees
Revenue—
Valuation$120Mestimated
Last round$12.2Mled by Felicis Ventures · 2023
Sources · 1 ↓
Suggest an editLast verified 2026-07-24