Pinned
The first-ever measured silicon numbers for @nvidia Vera Rubin NVL72 are in 😲
First measured performance shows 10x more tokens per megawatt than Blackwell.
No projections. Real results from live hardware.
CoreWeave
@CoreWeave
The Essential Cloud for AI
- Inference GPUs often sit idle overnight, when training needs the capacity. Join us August 18 for episode 3 of Training Tuesdays to see how SUNK Pod Scheduler runs training and inference on one cluster, helping teams use existing GPUs more effectively. utm.io/ur2sO
- Global trading firm IMC is expanding its footprint on CoreWeave to scale its next-gen quantitative research infrastructure. With immediately available, high-throughput compute, IMC researchers can process massive datasets and run parallel experiments faster without
- The question isn't what a token costs. It's how much of every GPU-hour is doing useful work. EVP @GoldbergChen on compute-bound vs. memory-bound workloads, and why benchmarks are the floor, not the ceiling in @CIOdive.
- The rumors are true. DeepSeek V4-Flash-0731 is LIVE on our Serverless Inference platform. You can put our GPUs to work here: utm.io/ur04rHey, I beat the @wandb intern to it: DeepSeek V4 is now live on serverless inference. Let’s see what they cooked.
