--- title: NVIDIA Nemotron 3.5 Lightning and NeMo Switchyard... description: NVIDIA has released Nemotron 3.5 Lightning, a 30-billion-parameter mixture-of-experts open model optimized for high-volume agentic AI tasks. It delivers up to... canonical: https://daily.dev/posts/nvidia-nemotron-3-5-lightning-and-nemo-switchyard-deliver-faster-smarter-more-efficient-agentic-ai-gclcvthzg twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: NVIDIA Nemotron 3.5 Lightning and NeMo Switchyard Deliver Faster, Smarter, More Efficient Agentic AI | daily.dev og:description: NVIDIA has released Nemotron 3.5 Lightning, a 30-billion-parameter mixture-of-experts open model optimized for high-volume agentic AI tasks. It delivers up to... og:url: https://daily.dev/posts/nvidia-nemotron-3-5-lightning-and-nemo-switchyard-deliver-faster-smarter-more-efficient-agentic-ai-gclcvthzg og:image: https://api.daily.dev/og/posts/GClCVthzG.png og:image:alt: NVIDIA Nemotron 3.5 Lightning and NeMo Switchyard Deliver Faster, Smarter, More Efficient 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. # NVIDIA Nemotron 3.5 Lightning and NeMo Switchyard Deliver Faster, Smarter, More Efficient Agentic AI **[NVIDIA](https://daily.dev/sources/nvidia)** · 6 min read · 1 upvotes · 0 comments ## Summary NVIDIA has released Nemotron 3.5 Lightning, a 30-billion-parameter mixture-of-experts open model optimized for high-volume agentic AI tasks. It delivers up to 4x faster output speed and 30% faster agentic task completion compared to similar models. Alongside it, NVIDIA released NeMo Switchyard, an open source model routing library that automatically directs prompts to the most suitable model in a multi-model system. Partner benchmarks show significant cost reductions: LangChain achieved 74% lower cost, Ramp cut costs by 58% while matching frontier model performance, and Cognition reduced mean cost by 28% using Switchyard in Devin Desktop. Nemotron 3.5 Lightning runs across RTX PCs, DGX Spark, DGX Station, Jetson, and cloud environments, and is available on Hugging Face, ModelScope, OpenRouter, and build.nvidia.com. NeMo Switchyard is available on GitHub. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is NVIDIA Nemotron 3.5 Lightning and what is it optimized for? Nemotron 3.5 Lightning is a 30-billion-parameter mixture-of-experts open model from NVIDIA, released as part of the Nemotron 3 family, built for high-volume specialized tasks within larger multi-agent systems such as code review, tool use, and billing questions. It delivers up to 4x faster output speed and 30% faster agentic task completion than comparable models, and can be post-trained with NVIDIA NeMo on domain-specific data. _Developers evaluating specialized agent models can track new Nemotron releases and benchmarks on daily.dev._ ### What is NVIDIA NeMo Switchyard and how much can it reduce AI agent costs? NeMo Switchyard is an open source model-routing library that automatically directs each agent request to the most capable and cost-efficient model for that step, without requiring code rewrites. Internal benchmarks show it maintains frontier-level accuracy while cutting task completion cost to nearly one-third of using Opus 4.8 alone; partners like Ramp and LangChain reported cost cuts of 58% and 74% respectively in their own workloads. _Teams weighing model routing strategies for agent cost control can follow real integration results on daily.dev._ ### Which companies have integrated or benchmarked NVIDIA NeMo Switchyard for agent routing? Partners including Boomi, Cadence, Classmethod, Cognition, Kong, LangChain, LiteLLM, Nous Research, Ramp, and Siemens have integrated or benchmarked NeMo Switchyard. Results vary: Boomi achieved 100% domain-routing accuracy while sending 59% of traffic to a faster fine-tuned model, Cognition cut mean cost 28% in Devin Desktop, and Ramp cut costs 58% and runtime 33% on Ramp SWE-Bench. _Anyone comparing agent routing integrations across vendors can follow these rollouts on daily.dev._ ## Community take How the wider developer community reacted, aggregated from 1 discussion and 63 comments across hackernews (as of 2026-09-14). **TL;DR:** The community is broadly interested in the trend toward smaller, efficient models that Nemotron 3.5 Lightning represents, but discussion quickly diverges into debates about whether small models are the future or just a stopgap, and there is notable skepticism about NeMo Switchyard's prompt-caching compatibility and its 'experimental, not for production' status. **Sentiment:** 30% positive · 40% mixed · 30% skeptical **The case for** - Smaller, efficient models enable qualitatively different user experiences beyond just being cheaper versions of frontier models. - The model runs well on Apple Silicon via MLX, making it accessible on consumer hardware. - Speed gains from efficient models can unlock new categories of applications, not just cost savings. **The pushback** - NeMo Switchyard's README explicitly labels it 'experimental, not for production use,' contradicting the press release's deployment messaging. - Prompt/KV caching is model-specific and breaks when switching between models, raising real questions about the router's practical cost savings. - Competing models like Meta's 30B dense offering may outperform Nemotron Lightning on quality benchmarks, and Qwen updates are imminent. - The benchmark graphs omit the Qwen model family (except an out-of-range variant), which commenters called out as selectively dishonest. - Some argue that scaling large models and distilling down is still the dominant path, and betting on small models alone is a 'cope' by labs without compute. **By community** - hackernews (mixed): Engaged but divided — optimistic about the small/efficient model trend in principle, yet skeptical of Switchyard's production readiness, caching claims, and benchmark honesty. **Hottest debate:** Whether small, efficient models represent a genuine architectural leap or are merely a temporary workaround until hardware and large-model scaling catch up. **Open questions** - How does NeMo Switchyard handle KV/prompt cache continuity when routing a session across different models? - What is the minimum VRAM required to run Nemotron 3.5 Lightning at Q4 quantization on consumer hardware? - Why does the README say 'not for production use' while the press release encourages deployment? - How does Nemotron 3.5 Lightning compare fairly against the full Qwen 3 family at similar active parameter counts? **Highlights** > The repo is probably a better entrypoint to it, bit more concise description than the press releases: https://github.com/NVIDIA-NeMo/Switchyard (Notably: "Experimental software. Not for production use."). Unclear if they actually want you to deploy it or not, press release says yes, README says no, do with that what you will. Doesn't seem to mention "cache" in the README nor the docs, but the code has mentions of it (https://github.com/search?q=repo%3ANVIDIA-NeMo%2FSwitchyard+...), I'm not sure what their thinking is there. "Good luck" essentially? Seems to be per-provider at best, but weird position for a routing library to take. > — [embedding-shape on hackernews · 1 comments](https://news.ycombinator.com/item?id=49263563) > To elaborate, because I don't think some of the people reading this understand the reason, typically a lot or most of the cost in "agentic" API usage is cached read + generation. Cached read costs scale with turn count, which multi-model switching doesn't increase, and of course generation gets cheaper if you do some of it with a cheaper model. When you switch models the "catching up" batch of messages is just a single prefill and then that goes into cache. You don't even need to have the same chat history across models so long as the view from each model's perspective looks like a series of appends. The main problem with model routing in my experience is that to work well the router needs to be pretty strong, maybe even moreso than any of the actual models in service. There are probably clever solutions to this but I haven't seen any that look better than just using sub-agents. > — [hedgehog on hackernews · 1 comments](https://news.ycombinator.com/item?id=49263806) > They conveniently decided not to include the Qwen range of models in the Artificial Analysis graph, except the out-of-league Max variant. At least be brave and honest. > — [docheinestages on hackernews](https://news.ycombinator.com/item?id=49264756) > Not by the looks of it, but it got me thinking, currently in the middle of Level1Techs coverage on the model and switchyard and he mentions "how easy it is to customize it". Fully admitting that I haven't yet read the docs, my issue with that is "we can train LORAs for 35B as well, why use this (according to benchmarks) worse model for customization instead of a slightly bigger better one?" Assuming I eat my words after going through the docs and this is actually a more efficient model / loras adapt better, I don't see as much value in it as is, as a REAP of it (remove least-important experts, domain-locked tests show ~98% retained accuracy) to something like 20B-A3B (rouhgly matching gpt oss, which while a good model, is outdated knowledge-wise and not as good with tool in my xp). Having a 20B-A3B model at q4 that has a lora to be your local orchestrator (delegating coding to server/cloud models) and ci/cd runner does start sounding like an appealing proposition to me, as that would fit in 16gb vram easily (fitting many consumer gpus and 24gb macs). > — [sleepyeldrazi on hackernews](https://news.ycombinator.com/item?id=49265159) > I think finding significant efficiency gains with LLMs and the like may lead to qualitatively better products. Looking at people's experiences to DSV4F makes me believe that even more than before too. I don't think people are realizing that speed can allow for categorically different user experiences that are more than just "worse than frontier capabilities but faster". > — [cootsnuck on hackernews](https://news.ycombinator.com/item?id=49264987) **Source threads** - [hackernews](https://news.ycombinator.com/item?id=49263340) · 124 points · 63 comments ## Similar posts on daily.dev - [Nvidia launches a smaller, faster Nemotron model and a router to put it to work](https://daily.dev/posts/nvidia-launches-a-smaller-faster-nemotron-model-and-a-router-to-put-it-to-work-adgtw6klw) · The New Stack · 0 upvotes · 0 comments - [NVIDIA Nemotron 3.5 Lightning · Ollama Blog](https://daily.dev/posts/nvidia-nemotron-3-5-lightning-ollama-blog-uy1kba1ur) · Ollama · 3 upvotes · 0 comments - [The Fastest Nemotron Yet: Embracing NVIDIA Nemotron 3.5 Lightning in Kilo](https://daily.dev/posts/the-fastest-nemotron-yet-embracing-nvidia-nemotron-3-5-lightning-in-kilo-wcpgec5ll) · Kilo Blog · 4 upvotes · 2 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#nvidia](https://daily.dev/tags/nvidia), [#agentic-ai](https://daily.dev/tags/agentic-ai), [#mixture-of-experts](https://daily.dev/tags/mixture-of-experts), [#ai-gateway](https://daily.dev/tags/ai-gateway) [View this post on daily.dev](https://daily.dev/posts/nvidia-nemotron-3-5-lightning-and-nemo-switchyard-deliver-faster-smarter-more-efficient-agentic-ai-gclcvthzg) ```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":"NVIDIA Nemotron 3.5 Lightning and NeMo Switchyard Deliver Faster, Smarter, More Efficient Agentic AI","url":"https://daily.dev/posts/nvidia-nemotron-3-5-lightning-and-nemo-switchyard-deliver-faster-smarter-more-efficient-agentic-ai-gclcvthzg","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/nvidia-nemotron-3-5-lightning-and-nemo-switchyard-deliver-faster-smarter-more-efficient-agentic-ai-gclcvthzg"},"datePublished":"2026-08-11T13:07:24.434Z","dateModified":"2026-09-14T06:55:03.004Z","description":"NVIDIA has released Nemotron 3.5 Lightning, a 30-billion-parameter mixture-of-experts open model optimized for high-volume agentic AI tasks. It delivers up to...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/55c8a50d5c7b97047da9d72d32cdb6a8?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/55c8a50d5c7b97047da9d72d32cdb6a8?_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/nvidia-nemotron-3-5-lightning-and-nemo-switchyard-deliver-faster-smarter-more-efficient-agentic-ai-gclcvthzg","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,nvidia,agentic-ai,mixture-of-experts,ai-gateway","timeRequired":"PT6M"} {"@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":"NVIDIA Nemotron 3.5 Lightning and NeMo Switchyard Deliver Faster, Smarter, More Efficient Agentic AI"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/nvidia-nemotron-3-5-lightning-and-nemo-switchyard-deliver-faster-smarter-more-efficient-agentic-ai-gclcvthzg#faq","mainEntity":[{"@type":"Question","name":"What is NVIDIA Nemotron 3.5 Lightning and what is it optimized for?","acceptedAnswer":{"@type":"Answer","text":"Nemotron 3.5 Lightning is a 30-billion-parameter mixture-of-experts open model from NVIDIA, released as part of the Nemotron 3 family, built for high-volume specialized tasks within larger multi-agent systems such as code review, tool use, and billing questions. It delivers up to 4x faster output speed and 30% faster agentic task completion than comparable models, and can be post-trained with NVIDIA NeMo on domain-specific data. Developers evaluating specialized agent models can track new Nemotron releases and benchmarks on daily.dev."}},{"@type":"Question","name":"What is NVIDIA NeMo Switchyard and how much can it reduce AI agent costs?","acceptedAnswer":{"@type":"Answer","text":"NeMo Switchyard is an open source model-routing library that automatically directs each agent request to the most capable and cost-efficient model for that step, without requiring code rewrites. Internal benchmarks show it maintains frontier-level accuracy while cutting task completion cost to nearly one-third of using Opus 4.8 alone; partners like Ramp and LangChain reported cost cuts of 58% and 74% respectively in their own workloads. Teams weighing model routing strategies for agent cost control can follow real integration results on daily.dev."}},{"@type":"Question","name":"Which companies have integrated or benchmarked NVIDIA NeMo Switchyard for agent routing?","acceptedAnswer":{"@type":"Answer","text":"Partners including Boomi, Cadence, Classmethod, Cognition, Kong, LangChain, LiteLLM, Nous Research, Ramp, and Siemens have integrated or benchmarked NeMo Switchyard. Results vary: Boomi achieved 100% domain-routing accuracy while sending 59% of traffic to a faster fine-tuned model, Cognition cut mean cost 28% in Devin Desktop, and Ramp cut costs 58% and runtime 33% on Ramp SWE-Bench. Anyone comparing agent routing integrations across vendors can follow these rollouts on daily.dev."}}]} ```