--- title: A used Tesla V100 has quietly become the cheapest way to... description: The Nvidia Tesla V100 (PCIe, 32GB) has emerged as a budget option for running local LLMs at home, often undercutting used RTX 3090 prices while offering 8GB... canonical: https://daily.dev/posts/a-used-tesla-v100-has-quietly-become-the-cheapest-way-to-run-local-llms-at-home-ku98vsxlo twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: A used Tesla V100 has quietly become the cheapest way to run local LLMs at home | daily.dev og:description: The Nvidia Tesla V100 (PCIe, 32GB) has emerged as a budget option for running local LLMs at home, often undercutting used RTX 3090 prices while offering 8GB... og:url: https://daily.dev/posts/a-used-tesla-v100-has-quietly-become-the-cheapest-way-to-run-local-llms-at-home-ku98vsxlo og:image: https://api.daily.dev/og/posts/ku98VSxlo.png og:image:alt: A used Tesla V100 has quietly become the cheapest way to run local LLMs at home 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. # A used Tesla V100 has quietly become the cheapest way to run local LLMs at home **[XDA Developers](https://daily.dev/sources/xda-developers)** · 5 min read · 1 upvotes · 2 comments ## Summary The Nvidia Tesla V100 (PCIe, 32GB) has emerged as a budget option for running local LLMs at home, often undercutting used RTX 3090 prices while offering 8GB more VRAM. The 32GB model starts around $400–$500 from overseas sellers versus $700–$1,050 for a used 3090. However, significant caveats exist: CUDA Toolkit 13.0 dropped Volta support, PyTorch is moving to deprecate compute capability 7.0, LM Studio doesn't support it, and prebuilt llama.cpp binaries no longer include Volta — requiring manual compilation. The card also lacks BF16 and FP8 support, runs passively cooled (requiring airflow mods for ATX cases), and has no display outputs. For those who want a hassle-free inference setup, the RTX 3090 remains the safer choice, but the V100 32GB is the cheapest path past the 24GB VRAM ceiling for those willing to work around its limitations. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Does CUDA Toolkit 13.0 still support the Nvidia Tesla V100 GPU? No, CUDA Toolkit 13.0 removed offline compilation and library support for Maxwell, Pascal, and Volta architectures, which includes the Tesla V100. Applications built with the CUDA 12.x series continue to work, but newer toolkits cannot target these architectures at all, and CUDA 13.3 release notes confirm support for pre-Turing architectures has been dropped. _Track GPU driver and toolkit deprecations like this before betting your local LLM setup on older hardware, with daily.dev._ ### Is a used Tesla V100 32GB cheaper than a used RTX 3090 for running local LLMs? Yes, at the low end a 32GB Tesla V100 costs roughly half the price of a used RTX 3090 while offering a third more memory. As of late July 2026, 32GB V100 cards start around $400-500 from overseas sellers versus $700-1,050 for a used RTX 3090, though refurbished V100 units with a return window run closer to $880-900, narrowing the gap at the high end. _Compare GPU price and capability tradeoffs like V100 versus RTX 3090 before buying hardware for local LLM inference on daily.dev._ ### Does LM Studio support the Nvidia Tesla V100 GPU for local LLM inference? No, LM Studio's CUDA runtime does not support the Tesla V100. Additionally, prebuilt llama.cpp binaries have dropped Volta support, meaning V100 owners need to compile llama.cpp themselves to use it, and PyTorch has an open proposal to deprecate Volta support outright after moving compute capability 7.0 users onto the CUDA 12.6 wheel. _Keep tabs on which inference tools still support older GPU architectures before committing to a local LLM setup, via daily.dev._ ## Community discussion Top comments from developers on daily.dev. **@robertkeogh** · 0 upvotes > Maybe its still good for Stable Diffusion **@johnstewart62** · 0 upvotes > I thought for a second you were talking about Tesla the car 😬 ## Similar posts on daily.dev - [I almost bought a used Nvidia Tesla GPU for my home lab, then I read what owners actually deal with](https://daily.dev/posts/i-almost-bought-a-used-nvidia-tesla-gpu-for-my-home-lab-then-i-read-what-owners-actually-deal-with-r63cwpg1x) · XDA Developers · 2 upvotes · 0 comments - [Getting A Proprietary-Bus GPU Onto PCIe Enables Cheaper Local LLMs, For Now](https://daily.dev/posts/getting-a-proprietary-bus-gpu-onto-pcie-enables-cheaper-local-llms-for-now-9vqydfswh) · Hackaday · 0 upvotes · 0 comments - [I added a second GPU just for local AI workloads, and it cost less than upgrading my main one](https://daily.dev/posts/i-added-a-second-gpu-just-for-local-ai-workloads-and-it-cost-less-than-upgrading-my-main-one-rnty4xi8p) · XDA Developers · 0 upvotes · 0 comments --- Tags: [#nvidia](https://daily.dev/tags/nvidia), [#gpu](https://daily.dev/tags/gpu), [#local-ai](https://daily.dev/tags/local-ai), [#cuda](https://daily.dev/tags/cuda), [#llama-cpp](https://daily.dev/tags/llama-cpp) [View this post on daily.dev](https://daily.dev/posts/a-used-tesla-v100-has-quietly-become-the-cheapest-way-to-run-local-llms-at-home-ku98vsxlo) ```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":"A used Tesla V100 has quietly become the cheapest way to run local LLMs at home","url":"https://daily.dev/posts/a-used-tesla-v100-has-quietly-become-the-cheapest-way-to-run-local-llms-at-home-ku98vsxlo","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/a-used-tesla-v100-has-quietly-become-the-cheapest-way-to-run-local-llms-at-home-ku98vsxlo"},"datePublished":"2026-07-28T16:01:13.953Z","dateModified":"2026-09-14T07:28:40.732Z","description":"The Nvidia Tesla V100 (PCIe, 32GB) has emerged as a budget option for running local LLMs at home, often undercutting used RTX 3090 prices while offering 8GB...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/c6d28064fd022db234610c6e63eb657e?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/c6d28064fd022db234610c6e63eb657e?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"XDA Developers","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":"XDA Developers","logo":"https://media.daily.dev/image/upload/s--kCg6yyAP--/f_auto,q_auto/v1774964407/logos/xda-developers?_a=BAMAMiWQ0","url":"https://daily.dev/sources/xda-developers"},"commentCount":2,"discussionUrl":"https://daily.dev/posts/a-used-tesla-v100-has-quietly-become-the-cheapest-way-to-run-local-llms-at-home-ku98vsxlo","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":2}],"keywords":"nvidia,gpu,local-ai,cuda,llama-cpp","timeRequired":"PT5M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"XDA Developers","item":"https://daily.dev/sources/xda-developers"},{"@type":"ListItem","position":3,"name":"A used Tesla V100 has quietly become the cheapest way to run local LLMs at home"}]} {"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/a-used-tesla-v100-has-quietly-become-the-cheapest-way-to-run-local-llms-at-home-ku98vsxlo","comment":[{"@type":"Comment","text":"Maybe its still good for Stable Diffusion","datePublished":"2026-07-29T21:00:05.343Z","url":"https://daily.dev/posts/ku98VSxlo#c-JcyvWIJU0","author":{"@type":"Person","name":"Robert keogh","url":"https://daily.dev/robertkeogh","image":"https://lh3.googleusercontent.com/a/ACg8ocJBETp-idwWWNDb2-Z6sGMIblVaZbAKctQqm3m825kihc1QZg=s96-c"}},{"@type":"Comment","text":"I thought for a second you were talking about Tesla the car 😬","datePublished":"2026-07-28T16:45:38.386Z","url":"https://daily.dev/posts/ku98VSxlo#c-1uPtsO5mA","author":{"@type":"Person","name":"John Stewart","url":"https://daily.dev/johnstewart62","image":"https://media.daily.dev/image/upload/s--kvHAbexP--/f_auto/v1783956120/avatars/avatar_4raeU5ZwsqH1K0EuYna8u?_a=BAMAMicg0"}}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/a-used-tesla-v100-has-quietly-become-the-cheapest-way-to-run-local-llms-at-home-ku98vsxlo#faq","mainEntity":[{"@type":"Question","name":"Does CUDA Toolkit 13.0 still support the Nvidia Tesla V100 GPU?","acceptedAnswer":{"@type":"Answer","text":"No, CUDA Toolkit 13.0 removed offline compilation and library support for Maxwell, Pascal, and Volta architectures, which includes the Tesla V100. Applications built with the CUDA 12.x series continue to work, but newer toolkits cannot target these architectures at all, and CUDA 13.3 release notes confirm support for pre-Turing architectures has been dropped. Track GPU driver and toolkit deprecations like this before betting your local LLM setup on older hardware, with daily.dev."}},{"@type":"Question","name":"Is a used Tesla V100 32GB cheaper than a used RTX 3090 for running local LLMs?","acceptedAnswer":{"@type":"Answer","text":"Yes, at the low end a 32GB Tesla V100 costs roughly half the price of a used RTX 3090 while offering a third more memory. As of late July 2026, 32GB V100 cards start around $400-500 from overseas sellers versus $700-1,050 for a used RTX 3090, though refurbished V100 units with a return window run closer to $880-900, narrowing the gap at the high end. Compare GPU price and capability tradeoffs like V100 versus RTX 3090 before buying hardware for local LLM inference on daily.dev."}},{"@type":"Question","name":"Does LM Studio support the Nvidia Tesla V100 GPU for local LLM inference?","acceptedAnswer":{"@type":"Answer","text":"No, LM Studio's CUDA runtime does not support the Tesla V100. Additionally, prebuilt llama.cpp binaries have dropped Volta support, meaning V100 owners need to compile llama.cpp themselves to use it, and PyTorch has an open proposal to deprecate Volta support outright after moving compute capability 7.0 users onto the CUDA 12.6 wheel. Keep tabs on which inference tools still support older GPU architectures before committing to a local LLM setup, via daily.dev."}}]} ```