--- title: From Training to Production, NVIDIA and CoreWeave Close... description: CoreWeave announced availability of NVIDIA Vera Rubin NVL72 systems with Spectrum-X networking, with Cognition (maker of the Devin AI software engineer) as the... canonical: https://daily.dev/posts/from-training-to-production-nvidia-and-coreweave-close-the-loop-on-agentic-ai-dhllfozzn twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI | daily.dev og:description: CoreWeave announced availability of NVIDIA Vera Rubin NVL72 systems with Spectrum-X networking, with Cognition (maker of the Devin AI software engineer) as the... og:url: https://daily.dev/posts/from-training-to-production-nvidia-and-coreweave-close-the-loop-on-agentic-ai-dhllfozzn og:image: https://api.daily.dev/og/posts/dhlLfOzZN.png og:image:alt: From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI **[NVIDIA](https://daily.dev/sources/nvidia)** · 7 min read · 1 upvotes · 0 comments ## Summary CoreWeave announced availability of NVIDIA Vera Rubin NVL72 systems with Spectrum-X networking, with Cognition (maker of the Devin AI software engineer) as the first production customer, reporting up to 4.8x higher token throughput versus GB200 NVL72 on agentic coding workloads. CoreWeave also introduced NVIDIA Vera CPU, showing over 3x faster agent sandbox startup times, and launched CoreWeave Forge, a unified environment combining Weights & Biases, OpenPipe, and marimo for continuous training-to-production model improvement, featuring new tools like CoreWeave ARIA, Agent Lens, and Sandboxes, plus serverless fine-tuning and RL offerings now used by Canva, Capital One and MasterClass. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How much faster is NVIDIA Vera Rubin NVL72 than GB200 NVL72 for agentic coding inference? Vera Rubin NVL72 delivered up to a 4.8x increase in total token throughput over GB200 NVL72 for SWE-2 inference workloads, based on early benchmarks by Cognition using a real-world software engineering workload sampled from FrontierCode tasks solved by AI agents. The gains translate into faster real-time code generation and more responsive multistep reasoning for Cognition's Devin AI software engineer. _Teams evaluating GPU platforms for agentic coding workloads can track throughput benchmarks like this one on daily.dev._ ### What performance improvement does NVIDIA Vera CPU offer for agent sandbox startup times? NVIDIA Vera CPU delivered more than 3x faster agent sandbox startup times in CoreWeave testing, along with a 1.7x performance gain on Terminal-Bench across all passing tasks. CoreWeave's deployment puts 128 CPUs and 11,264 cores in a single rack, supporting more than 11,000 concurrent hardware-isolated agent environments for reinforcement learning and tool use. _Anyone scaling isolated agent sandboxes for RL or tool use can follow CPU benchmark results like these on daily.dev._ ### What is CoreWeave Forge and what tools does it combine? CoreWeave Forge is a connected environment for training, evaluating, and improving models and agents that unifies Weights & Biases, post-training expertise from OpenPipe, and the open source marimo notebook project into one platform. It includes CoreWeave ARIA for experiment analysis, Agent Lens for production observability, Sandboxes for isolated execution, and serverless supervised fine-tuning and RL that trains 1.4x faster at 40% lower cost than self-managed setups. _Developers piecing together training and observability tools for agent improvement loops can follow platform launches like this on daily.dev._ ## Similar posts on daily.dev - [NVIDIA Vera Rubin and Blackwell Set a New Standard for Agentic AI Performance per Watt](https://daily.dev/posts/nvidia-vera-rubin-and-blackwell-set-a-new-standard-for-agentic-ai-performance-per-watt-r1iqkqtkn) · NVIDIA Developer · 0 upvotes · 0 comments - [Nvidia invests $2B to help debt-ridden CoreWeave add 5GW of AI compute](https://daily.dev/posts/nvidia-invests-2b-to-help-debt-ridden-coreweave-add-5gw-of-ai-compute-suyuzcb2u) · TechCrunch · 0 upvotes · 0 comments - [Introducing CoreWeave ARENA, New Lab for AI Production Readiness](https://daily.dev/posts/introducing-coreweave-arena-new-lab-for-ai-production-readiness-pqxx9xpj1) · SD Times · 0 upvotes · 0 comments - [NVIDIA Launches Vera CPU, Purpose-Built for Agentic AI](https://daily.dev/posts/nvidia-launches-vera-cpu-purpose-built-for-agentic-ai-vcgvkme6v) · Hacker News · 1 upvotes · 0 comments --- Tags: [#nvidia](https://daily.dev/tags/nvidia), [#reinforcement-learning](https://daily.dev/tags/reinforcement-learning), [#agentic-ai](https://daily.dev/tags/agentic-ai) [View this post on daily.dev](https://daily.dev/posts/from-training-to-production-nvidia-and-coreweave-close-the-loop-on-agentic-ai-dhllfozzn) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI","url":"https://daily.dev/posts/from-training-to-production-nvidia-and-coreweave-close-the-loop-on-agentic-ai-dhllfozzn","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/from-training-to-production-nvidia-and-coreweave-close-the-loop-on-agentic-ai-dhllfozzn"},"datePublished":"2026-09-30T18:56:06.987Z","dateModified":"2026-09-30T18:56:32.465Z","description":"CoreWeave announced availability of NVIDIA Vera Rubin NVL72 systems with Spectrum-X networking, with Cognition (maker of the Devin AI software engineer) as the...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/edcb3999c7f59c98c4aea7cfbdf076a2?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/edcb3999c7f59c98c4aea7cfbdf076a2?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"NVIDIA","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"NVIDIA","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/236a5c1f410e450b8d696cf3a4bbe374","url":"https://daily.dev/sources/nvidia"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/from-training-to-production-nvidia-and-coreweave-close-the-loop-on-agentic-ai-dhllfozzn","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"nvidia,reinforcement-learning,agentic-ai","timeRequired":"PT7M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"NVIDIA","item":"https://daily.dev/sources/nvidia"},{"@type":"ListItem","position":3,"name":"From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/from-training-to-production-nvidia-and-coreweave-close-the-loop-on-agentic-ai-dhllfozzn#faq","mainEntity":[{"@type":"Question","name":"How much faster is NVIDIA Vera Rubin NVL72 than GB200 NVL72 for agentic coding inference?","acceptedAnswer":{"@type":"Answer","text":"Vera Rubin NVL72 delivered up to a 4.8x increase in total token throughput over GB200 NVL72 for SWE-2 inference workloads, based on early benchmarks by Cognition using a real-world software engineering workload sampled from FrontierCode tasks solved by AI agents. The gains translate into faster real-time code generation and more responsive multistep reasoning for Cognition's Devin AI software engineer. Teams evaluating GPU platforms for agentic coding workloads can track throughput benchmarks like this one on daily.dev."}},{"@type":"Question","name":"What performance improvement does NVIDIA Vera CPU offer for agent sandbox startup times?","acceptedAnswer":{"@type":"Answer","text":"NVIDIA Vera CPU delivered more than 3x faster agent sandbox startup times in CoreWeave testing, along with a 1.7x performance gain on Terminal-Bench across all passing tasks. CoreWeave's deployment puts 128 CPUs and 11,264 cores in a single rack, supporting more than 11,000 concurrent hardware-isolated agent environments for reinforcement learning and tool use. Anyone scaling isolated agent sandboxes for RL or tool use can follow CPU benchmark results like these on daily.dev."}},{"@type":"Question","name":"What is CoreWeave Forge and what tools does it combine?","acceptedAnswer":{"@type":"Answer","text":"CoreWeave Forge is a connected environment for training, evaluating, and improving models and agents that unifies Weights & Biases, post-training expertise from OpenPipe, and the open source marimo notebook project into one platform. It includes CoreWeave ARIA for experiment analysis, Agent Lens for production observability, Sandboxes for isolated execution, and serverless supervised fine-tuning and RL that trains 1.4x faster at 40% lower cost than self-managed setups. Developers piecing together training and observability tools for agent improvement loops can follow platform launches like this on daily.dev."}}]} ```