Lead story
Ask your AI
Top stories
Models & availability
Latest
← Feed
Copy✕ Close copy menu
60x Faster Cold Starts: Treating Peer GPUs as Weight Servers
Runway built a system called NCCLBack that reduces model cold-start times from minutes to seconds by broadcasting weights between GPUs instead of downloading them from storage.
From the source
How we cut model cold-start times from minutes to seconds by broadcasting weights between GPUs instead of downloading them from storage, and the sharp edges we found along the way.
runway.com
Visit source ↗runway.com
- Announced
- 04 May 2026
- Detected
- 03 Sep 2026 14:37 UTC (+121d after)