Cognition First to Run NVIDIA Vera Rubin on CoreWeave, ~4.8x Throughput vs GB200
silasalberti · x · 2026-10-01
- Cognition announced it is the first customer running NVIDIA Vera Rubin on CoreWeave, following Hopper (2022) and Blackwell (2024).
- Claimed numbers: on SWE-2 inference, Vera Rubin delivers about 4.8x more token throughput than GB200 at the same decode speed.
- The company ties the new hardware directly to agent quality: more compute, better agents.
Related event: Cognition First to Run NVIDIA Vera Rubin on CoreWeave(4 posts)→
More from coding & agent
- Developer live-builds an environment-aware agent harness plugin in public brainstorm — JnBrymn · 2026-10-01
- DSPy's Flex lets optimizers rewrite your pipeline code, not just prompts — lateinteraction · 2026-10-01
- Developer releases an Apple skill for Vision Pro UI — damienghader · 2026-10-01
- Architect's take: AI agents are making your app's frontend UI optional — rseroter · 2026-10-01
- "Opus 5.5 is the cheapest model" — human hours saved beat token pricing — CamBrazy3 · 2026-10-01
- growth.engineer launches as an open-source catalog of GTM workflows you can copy into any agent — TAbrodi · 2026-10-01