CoreWeave (@CoreWeave) / X

CoreWeave

1,577 posts

CoreWeave

@CoreWeave

The Essential Cloud for AI

United States

Joined March 2018

Pinned
CoreWeave
@CoreWeave
Jun 18
We just trained DeepSeek-V3 671B benchmark in 2 minutes. 671 billion parameters. 8,192 @nvidia Blackwell Ultra GPUs. A 2 minute time-to-train, the fastest DeepSeek-V3 run ever recorded in MLPerf®. Check out the final results. utm.io/uqDci
GIF
21K
CoreWeave reposted
NVIDIA AI Infrastructure
@NVIDIAAIInfra
9h
Up to 10x better inference per watt. One-tenth the cost per million tokens. This is what NVIDIA Vera Rubin NVL72 delivers. Join us for the virtual event on June 30 as Dion Harris, Senior Director of HPC and AI Hyperscale Infrastructure Solutions at NVIDIA, and Harsh Banwait,
2.5K
CoreWeave
@CoreWeave
8h
If infrastructure variance is slowing down your models, you are leaving money on the table. General-purpose clouds simply weren't built for the intense demands of AI-scale quant research. Dive into the new joint whitepaper from CoreWeave and @VAST_Data to learn how to accelerate
2K
CoreWeave
@CoreWeave
8h
We are the first cloud provider to bring up and validate @nvidia Vera Rubin NVL72. Not just the hardware. The full stack: software-defined liquid cooling (Valvey), unified rack control (Racky), CoreWeave Mission Control across the fleet. Read the blog: utm.io/uqEQP
00:00
17K
CoreWeave
@CoreWeave
Jun 25
Powered on does not equal useful work 🛑 While the industry measures "utilization" by the amount of time a machine is simply turned on, our platform uses SUNK-orchestrated clusters to deliver up to 96% goodput. That means 96% of the GPU time you pay for goes directly into actual
00:00
3K
CoreWeave
@CoreWeave
Jun 25
We ran the DeepSeek-V3 671B MLPerf training benchmark in 2.02 mins 671 billion parameters. 8,192 @nvidia Blackwell Ultra GPUs. An MLPerf benchmark completed in 2 minutes, the fastest DeepSeek-V3 run ever recorded in MLPerf®. Check out the final results. utm.io/uqDci
00:00
4.9K
CoreWeave
@CoreWeave
Jun 25
Agentic AI is rewriting what infrastructure has to do. On June 30 at 9:30 am PT, tune into @SiliconANGLE @theCUBE as we discuss and demo @nvidia Vera Rubin NVL72. Save the date. bit.ly/43Spi9L
00:00
3.1K
CoreWeave
@CoreWeave
Jun 24
High-performance compute, lower-priced tickets. Take 33% off your #FullyConnected26 pass with code OSCW33. Get your discounted tickets here. utm.io/uqJPI
00:00
2.3K
CoreWeave
@CoreWeave
Jun 24
Open weights just caught up to the frontier. GLM-5.2 from @Zai_org tops the open-model rankings on @ArtificialAnlys and @arena's Agent Arena. It's now live on CoreWeave Serverless Inference at $1.39 in and $4.40 out per 1M tokens. Ship more for less.
00:00
16K
CoreWeave
@CoreWeave
Jun 24
Run GLM-5.2 on CoreWeave Serverless Inference here: utm.io/uqTy4 See the engineering behind our speed: utm.io/uqTx0
Z.AI GLM 5.2
From wandb.ai
970
CoreWeave
@CoreWeave
Jun 24
At #FullyConnected26, every session is practitioner-first: real deployments, real infrastructure decisions, real outcomes. So, if you've built something worth sharing, tell us. Call for speakers closes June 26. bit.ly/4eWZyPJ
1.6K
CoreWeave
@CoreWeave
Jun 24
The arena just got smarter ⚙️ ICYMI: CoreWeave is now the Official AI Cloud Partner of the @BattleBots Pro League. We're giving competing teams access to our AI Cloud platform to simulate, test, and push what's possible. Building? Get the details. bit.ly/4aB0siz
00:00
3K
CoreWeave
@CoreWeave
Jun 23
CoreWeave joins @SiliconANGLE @theCUBE on June 30. Get an inside look of NVIDIA Vera Rubin NVL72 and the team behind the cloud bring-up. bit.ly/43Spi9L
00:00
2.8K
CoreWeave reposted
Weights & Biases
@wandb
Jun 23
Most teams training RL agents optimize for tokens per second. For RL, that's the wrong number to chase. So we rebuilt our backend around trajectories per second. 💥 Meet AOM, a Megatron backend for our open-source library ART, with 12X the throughput of our old Unsloth backend.
29K