River AI Raises $1.1B, Launches API for Personalized Model Training
marcbhargava · x · 2026-08-11
Startup River AI announced it has raised $1.1 billion, led by General Catalyst with participation from NVIDIA, AMD, and Y Combinator. The company aims to build a personal AI stack entirely owned by the user.
Their core product, River API, is now in preview, promoting a "stop prompting, start training" philosophy. Key features include:
- LoRA Fine-tuning & RL: Supports low-rank adapter fine-tuning and reinforcement learning on open-source models ranging from 35B to 1T parameters (e.g., Qwen, Kimi, GLM series).
- Weight Updates & Ownership: Unlike prompting, the service actually updates model weights. The resulting checkpoint is fully owned by the user, who can iterate and serve it in production.
- Pay-per-token: Billed on tokens for both inference and training (e.g., Qwen3.5-9B training is $1.46 / 1M tokens) without GPU-hour math.
Related event: River AI Raises $1.1B to Build Personalized AI(6 posts)→
More from Venture
- Open-Source Model Customization Startup Trajectory Raises New Funding at $300M Valuation — steph_palazzolo · 2026-08-11
- Use AI to find local high-ticket service opportunities: a scoring system to find blue oceans — boringmarketer · 2026-08-11
- AI Protein Design Goes Big: Profluent's Lilly Deal Targets Large-Scale Gene Edits — nathanbenaich · 2026-08-11
- MacPaw Co-founder: Built 15 Products with Zero VC by Investing in AI Over Headcount — rschmelzer · 2026-08-11
- Amex Ventures Invests in Fazeshift to Build the Intelligence Layer for Finance Teams — ycombinator · 2026-08-11
- Big Tech Capex Surge: Quarterly Bills to Exceed $10B Amidst AI Infrastructure Boom — Beth_Kindig · 2026-08-11