Zoom says three years of fine-tuned Llama helped it cut AI costs substantially
firstadopter · x · 2026-07-26
Most companies will likely end up using frontier models for the hardest tasks and cheaper models for the rest.
A WSJ quote in the post says executives who use Anthropic and OpenAI want the highest available intelligence and are willing to pay a premium for it. Zoom CTO Xuedong Huang says Zoom has used Meta’s open-weight Llama for three years and fine-tuned it to save substantially. Zoom currently combines Anthropic, OpenAI and open models, and Huang describes this mix as the company’s “secret sauce.”
More from Companies & People
- Cursor holds a Mumbai community roadshow and teases a Pune event next — IndraVahan · 2026-07-26
- Anthropic allocates $200 million to RCTs and policy pilots — NathanpmYoung · 2026-07-26
- A 100-app sheet maps which AI apps are spending on Meta and TikTok ads — eptwts · 2026-07-26
- Jensen Huang says open models are critical for defending against rogue AI agents — Borthwick · 2026-07-26
- DeepSeek reportedly pauses plans for a second funding round — Polymarket · 2026-07-26
- Cisco-backed post says open-source AI models create superlinear value for security — aminkarbasi · 2026-07-26