CoreWeave Inc. (Nasdaq: CRWV) announced the availability of Nvidia Vera Rubin NVL72 on its cloud platform, with AI company Cognition as the first customer running production workloads on the system, according to a company press release.

Cognition, the applied AI lab behind the coding agent Devin, set up a Vera Rubin NVL72 cluster with CoreWeave in early September. Cognition's engineers ran the first customer-executed Vera Rubin inference benchmark, measured against a GB200 NVL72 cluster baseline. The benchmarks recorded a 4.8 times increase in total token throughput for SWE-2 inference workloads and a 3.8 times boost in output token throughput for reinforcement learning workloads on the Vera Rubin NVL72 compared to the GB200 NVL72 baseline.

"Agentic coding is an unforgiving workload that requires long contexts, high concurrency and rapid reasoning," said Silas Alberti, SVP research and founding team at Cognition. "By deploying the Nvidia Vera Rubin NVL72 on CoreWeave, our engineers are seeing up to a 4.8 times increase in total token throughput for SWE-2 inference workloads."

Cognition scaled from bridge capacity to thousands of GPUs for training and inference in less than nine months on CoreWeave. Customers running Vera Rubin NVL72 operate under the same tooling as existing Nvidia GB200 NVL72 and GB300 NVL72 deployments.

CoreWeave also claims to have published the first measured silicon performance numbers on the platform, citing 10 times the token throughput per megawatt over Nvidia GB200 NVL72 on the DeepSeek R1 reasoning model.

"CoreWeave has consistently demonstrated the infrastructure expertise required to bring each new generation of Nvidia accelerated computing into production," said Ian Buck, vice president of Hyperscale and High-Performance Computing at Nvidia.

CoreWeave, in collaboration with Dell Technologies, was among the first cloud providers to deploy Dell PowerRack systems featuring Nvidia GB200 and GB300 NVL72. The announcement was made at Fully Connected, CoreWeave's AI cloud conference attended by more than 4,500 participants.