Lambda Labs Breaks Record with All-Reduce on 10,368 GB300 GPUs
TheZachMueller · x · 2026-08-22
Lambda Labs announced a new record for the largest cluster-wide All-Reduce on its platform, running across 10,368 NVIDIA GB300 GPUs. The test utilized 9 pods and 144 racks, marking a 2.5x increase over their previous record.
Related event: Lambda Runs All-Reduce Test Across Over 10,000 GB300 GPUs(2 posts)→
More from Infra
- Lium Offers On-Demand B300 and RTX 5090 GPU Rentals — markjeffrey · 2026-08-22
- Vercel cuts GPT-5.6 Sol prices further on AI Gateway — cramforce · 2026-08-22
- Nvidia partners with data center developer Cloverleaf — TechCrunch AI · 2026-08-22
- ComfyUI on Windows Gains Unified Memory Support — tostane · 2026-08-22
- Mac Idle Compute Network Darkbloom Hits $102K ARR — gajesh · 2026-08-22
- One-click script automates vLLM setup on dual 4090s — doodlestein · 2026-08-22