{"slug": "tinygrad", "name": "tinygrad", "category": "dev-libs", "category_name": "Libraries and frameworks", "description": "", "recommenders": 8, "commenters": 22, "warners": 1, "score": 9.541, "rising": -2.62, "first": "2025-11-19", "latest": "2026-08-16", "why": "Absolutely amazing framework and you should absolutely check them out.", "why_is_quote": true, "has_verdict": false, "sentiment": {"recommend": 9, "neutral": 13, "compare": 0, "warn": 1}, "spark": [1, 3, 0, 0, 0, 1, 2, 1, 1, 0, 0, 0], "listed": true, "loved": [], "caveat": "", "short_quote": "Absolutely amazing framework and you should absolutely check them out.", "kind": "", "icon": false, "domain": "tinygrad.org", "github_repo": null, "official_url": "https://tinygrad.org", "pricing": "", "aliases": ["tinygrad"], "verdict": null, "praise": [{"label": "", "summary": "", "quotes": [{"id": 786075152, "item_id": 49129697, "url": "https://news.ycombinator.com/item?id=49129697", "by": "coretx", "date": "2026-07-31", "stance": "recommend", "quote": "tinygrad.org is probably the best punk in this regard", "excerpt": "https://tinygrad.org is probably the best punk in this regard. Moving LLM's out of your drug dealers datacenter and into your utility closet.", "context": "", "use_case": "running LLMs locally", "switched_from": "", "story": "Twenty-five years ago it was cryptography, today it's model weights"}, {"id": 776274752, "item_id": 48517172, "url": "https://news.ycombinator.com/item?id=48517172", "by": "Avicebron", "date": "2026-06-13", "stance": "recommend", "quote": "I'm a firm supporter of local inference though so good on them for doing something", "excerpt": "https://tinygrad.org/#tinybox I'm not sure exactly why you would buy through them vs rolling your own if you could afford the equivalent hardware. I'm a firm supporter of local inference though so good on them for doing something", "context": "", "use_case": "", "switched_from": "", "story": "Open Source AI Must Win"}, {"id": 771769088, "item_id": 48235568, "url": "https://news.ycombinator.com/item?id=48235568", "by": "reedf1", "date": "2026-05-22", "stance": "recommend", "quote": "At least geohot knows what he is doing.", "excerpt": "Buy one of these next time, https://tinygrad.org/#tinybox. At least geohot knows what he is doing.", "context": "", "use_case": "", "switched_from": "", "story": "Was my $48K GPU server worth it?"}, {"id": 769787537, "item_id": 48111721, "url": "https://news.ycombinator.com/item?id=48111721", "by": "fluxin", "date": "2026-05-12", "stance": "recommend", "quote": "Absolutely amazing framework and you should absolutely check them out.", "excerpt": "Started as an experiment playing with the latest foundational models to see the capabilities and optimize my workflows, turns out to be pretty functional. Maintains upstream UOp parity throughout. Lots of AI slop involved and a surprising amount of hand scaffolding needed to make it work initially. Learned quite a bit of the internals of tinygrad along the way. Absolutely amazing framework and you should absolutely c", "context": "", "use_case": "ML framework internals", "switched_from": "", "story": null}, {"id": 762257008, "item_id": 47641063, "url": "https://news.ycombinator.com/item?id=47641063", "by": "arjie", "date": "2026-04-04", "stance": "recommend", "quote": "Woah, this is exciting.", "excerpt": "Woah, this is exciting. I'm traveling but I have a 5090 lying around at home. I'm eager to give it a go. Docs are here: https://docs.tinygrad.org/tinygpu/ I hope it'll work on an M4 Mac Mini. Does anyone know what hardware to get? You'll need a full ATX PSU to supply power, right? And then tinygrad can do LLM inference on it?", "context": "", "use_case": "GPU driver for eGPUs", "switched_from": "", "story": "Apple approves driver that lets Nvidia eGPUs work with Arm Macs"}]}], "caveats": [{"label": "", "summary": "", "quotes": [{"id": 742094976, "item_id": 46380936, "url": "https://news.ycombinator.com/item?id=46380936", "by": "refulgentis", "date": "2025-12-24", "stance": "warn", "quote": "I know George has pushed this idea for years now", "excerpt": "My deepest apologies, I can't parse this and I earnestly tried: 5 minutes of my own thinking, then 3 llms, then a 10 minute timer of my own thinking over the whole thing. My guess is you're trying to communicate \"tinygrad doesn't need gpu drivers\" which maybe is transmutated into \"tinygrad replaces CUDA\" and you think \"CUDA means other GPUs can't be used for LLMs, thus nvidia has a strangehold\" I know George has push", "context": "", "use_case": "", "switched_from": "", "story": "Nvidia buying AI chip startup Groq for about $20B in cash"}]}], "who": [], "tags": [], "use_cases": ["deep learning on egpu", "running llms locally", "ml framework internals", "egpu driver for nvidia cards", "gpu driver for egpus", "deep learning framework"], "compared": [{"slug": "pytorch", "name": "PyTorch", "co": 2, "switched_from": 0, "switched_to": 0}], "makers": [], "threads": [{"url": "https://news.ycombinator.com/item?id=47640380", "title": "Apple approves driver that lets Nvidia eGPUs work with Arm Macs", "mentions": 5}, {"url": "https://news.ycombinator.com/item?id=46422757", "title": "Five Years of Tinygrad", "mentions": 5}, {"url": "https://news.ycombinator.com/item?id=49292385", "title": "Chestnut \u2013 eGPU dock with open-source firmware", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49083599", "title": "Twenty-five years ago it was cryptography, today it's model weights", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49079454", "title": "PyTorch: A Reference Language", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48511908", "title": "Open Source AI Must Win", "mentions": 1}], "sentiment_by_month": {"2025-11": {"recommend": 1}, "2025-12": {"neutral": 5, "recommend": 3, "warn": 1}, "2026-01": {}, "2026-02": {}, "2026-03": {"neutral": 1}, "2026-04": {"recommend": 1, "neutral": 4}, "2026-05": {"recommend": 2, "neutral": 1}, "2026-06": {"recommend": 1}, "2026-07": {"neutral": 1, "recommend": 1}, "2026-08": {"neutral": 1}, "2026-09": {}, "2026-10": {}}, "quotes": [{"id": 789115760, "item_id": 49319735, "url": "https://news.ycombinator.com/item?id=49319735", "by": "rvz", "date": "2026-08-16", "stance": "neutral", "quote": "I know it just works well with tinygrad.", "excerpt": "Good, something that is needed for running LLMs and other non-LLM based deep learning work that doesn't suck or has poor drivers / firmware for. I can rest assured that when I use this on any machine running Mac, Windows or Linux, I know it just works well with tinygrad.", "context": "", "use_case": "deep learning on eGPU", "switched_from": "", "story": "Chestnut \u2013 eGPU dock with open-source firmware"}, {"id": 786075152, "item_id": 49129697, "url": "https://news.ycombinator.com/item?id=49129697", "by": "coretx", "date": "2026-07-31", "stance": "recommend", "quote": "tinygrad.org is probably the best punk in this regard", "excerpt": "https://tinygrad.org is probably the best punk in this regard. Moving LLM's out of your drug dealers datacenter and into your utility closet.", "context": "", "use_case": "running LLMs locally", "switched_from": "", "story": "Twenty-five years ago it was cryptography, today it's model weights"}, {"id": 785358864, "item_id": 49084929, "url": "https://news.ycombinator.com/item?id=49084929", "by": "ofou", "date": "2026-07-28", "stance": "neutral", "quote": "this is just https://github.com/tinygrad/tinygrad", "excerpt": "this is just https://github.com/tinygrad/tinygrad", "context": "", "use_case": "", "switched_from": "", "story": "PyTorch: A Reference Language"}, {"id": 776274752, "item_id": 48517172, "url": "https://news.ycombinator.com/item?id=48517172", "by": "Avicebron", "date": "2026-06-13", "stance": "recommend", "quote": "I'm a firm supporter of local inference though so good on them for doing something", "excerpt": "https://tinygrad.org/#tinybox I'm not sure exactly why you would buy through them vs rolling your own if you could afford the equivalent hardware. I'm a firm supporter of local inference though so good on them for doing something", "context": "", "use_case": "", "switched_from": "", "story": "Open Source AI Must Win"}, {"id": 771769088, "item_id": 48235568, "url": "https://news.ycombinator.com/item?id=48235568", "by": "reedf1", "date": "2026-05-22", "stance": "recommend", "quote": "At least geohot knows what he is doing.", "excerpt": "Buy one of these next time, https://tinygrad.org/#tinybox. At least geohot knows what he is doing.", "context": "", "use_case": "", "switched_from": "", "story": "Was my $48K GPU server worth it?"}, {"id": 770962913, "item_id": 48185182, "url": "https://news.ycombinator.com/item?id=48185182", "by": "zer0zzz", "date": "2026-05-18", "stance": "neutral", "quote": "Be surprised if there\u2019s more developments like the tinygrad driver", "excerpt": "> Apple is a fashion company. They don't care about improving usability of any of their devices. The most important thing is to them is selling a lifestyle. I don\u2019t think you read what I said, or you are intentionally being dense. My point is that users and the industry have largely moved on from the modular pc box concept. Those who cling to it do it out of a sense of identity. > They don't care about improving usab", "context": "", "use_case": "", "switched_from": "", "story": "RTX 5090 and M4 MacBook Air: Can It Game?"}, {"id": 769787537, "item_id": 48111721, "url": "https://news.ycombinator.com/item?id=48111721", "by": "fluxin", "date": "2026-05-12", "stance": "recommend", "quote": "Absolutely amazing framework and you should absolutely check them out.", "excerpt": "Started as an experiment playing with the latest foundational models to see the capabilities and optimize my workflows, turns out to be pretty functional. Maintains upstream UOp parity throughout. Lots of AI slop involved and a surprising amount of hand scaffolding needed to make it work initially. Learned quite a bit of the internals of tinygrad along the way. Absolutely amazing framework and you should absolutely c", "context": "", "use_case": "ML framework internals", "switched_from": "", "story": null}, {"id": 762759744, "item_id": 47672484, "url": "https://news.ycombinator.com/item?id=47672484", "by": "rldjbpin", "date": "2026-04-07", "stance": "neutral", "quote": "", "excerpt": "based on your card, it should be using a decent bit of the 600w or so passed through the new 16-pin connector. goes without saying, a proper PSU (doesn't have to be ATX, but at least 750W to be on the safer side) is a must. for thunderbolt enclosures, consider going through the list - https://egpu.io/best-egpu-buyers-guide/#tb3-enclosures zero idea about mac support so YMMV.", "context": "", "use_case": "eGPU driver for Nvidia cards", "switched_from": "", "story": "Apple approves driver that lets Nvidia eGPUs work with Arm Macs"}, {"id": 762309088, "item_id": 47644318, "url": "https://news.ycombinator.com/item?id=47644318", "by": "ece", "date": "2026-04-04", "stance": "neutral", "quote": "now runs tinygrad on eGPUs", "excerpt": "Apple should update this page for ARM macs, now runs tinygrad on eGPUs: https://support.apple.com/en-us/102363", "context": "", "use_case": "", "switched_from": "", "story": "Apple approves driver that lets Nvidia eGPUs work with Arm Macs"}, {"id": 762289712, "item_id": 47643107, "url": "https://news.ycombinator.com/item?id=47643107", "by": "syntaxing", "date": "2026-04-04", "stance": "neutral", "quote": "From what I understand, only works with Tinygrad.", "excerpt": "From what I understand, only works with Tinygrad. Which is better than nothing but CUDA or Vulkan on pytorch isn\u2019t going to work from this. [1] https://docs.tinygrad.org/tinygpu/", "context": "", "use_case": "", "switched_from": "", "story": "Apple approves driver that lets Nvidia eGPUs work with Arm Macs"}, {"id": 762265616, "item_id": 47641601, "url": "https://news.ycombinator.com/item?id=47641601", "by": "999900000999", "date": "2026-04-04", "stance": "neutral", "quote": "", "excerpt": "You can buy a cheap GPU enclosure for about 100$ off ali express. Takes a standard PSU. However, Mac Minis don't have occulink. So you might be a bit limited by whatever USB C can do. Now if Intel can get there Arc drivers in order we'll see some real budget fun. https://www.newegg.com/intel-arc-pro-b70-32gb-graphics-card/p/N82E16814883008?item=N82E16814883008 32 GB of VRAM for 1000$. Plus a 500$ Mac Mini.", "context": "", "use_case": "", "switched_from": "", "story": "Apple approves driver that lets Nvidia eGPUs work with Arm Macs"}, {"id": 762257008, "item_id": 47641063, "url": "https://news.ycombinator.com/item?id=47641063", "by": "arjie", "date": "2026-04-04", "stance": "recommend", "quote": "Woah, this is exciting.", "excerpt": "Woah, this is exciting. I'm traveling but I have a 5090 lying around at home. I'm eager to give it a go. Docs are here: https://docs.tinygrad.org/tinygpu/ I hope it'll work on an M4 Mac Mini. Does anyone know what hardware to get? You'll need a full ATX PSU to supply power, right? And then tinygrad can do LLM inference on it?", "context": "", "use_case": "GPU driver for eGPUs", "switched_from": "", "story": "Apple approves driver that lets Nvidia eGPUs work with Arm Macs"}, {"id": 761562241, "item_id": 47597640, "url": "https://news.ycombinator.com/item?id=47597640", "by": "h4ch1", "date": "2026-03-31", "stance": "neutral", "quote": "", "excerpt": "announcement tweet: https://xcancel.com/__tinygrad__/status/2039213719155310736", "context": "", "use_case": "", "switched_from": "", "story": "TinyGPU \u2013 Use AMD and Nvidia GPUs on macOS with Tinygrad"}, {"id": 743072464, "item_id": 46442029, "url": "https://news.ycombinator.com/item?id=46442029", "by": "arjie", "date": "2025-12-30", "stance": "recommend", "quote": "so I'm excited to see it work. Rooting for the guy.", "excerpt": "I remember when he launched everything and was hoping that the AMD hardware (very capable) was just hamstrung by software. The idea that the easy part is actually the chips is something I never considered though I \"know\" that \"Nvidia succeeded because of their software\". Haha, very clever. I've got a Comma 3X and I'm thankful for the 4090 P2P work too (which is now here? https://github.com/tinygrad/open-gpu-kernel-mo", "context": "", "use_case": "", "switched_from": "", "story": "Five Years of Tinygrad"}, {"id": 743070640, "item_id": 46441915, "url": "https://news.ycombinator.com/item?id=46441915", "by": "mellosouls", "date": "2025-12-30", "stance": "neutral", "quote": "While low line count is a guiding light of this project", "excerpt": "From the instructions for potential contributors: No code golf! While low line count is a guiding light of this project, anything that remotely looks like code golf will be closed. The true goal is reducing complexity and increasing readability, and deleting \\ns does nothing to help with that. https://github.com/tinygrad/tinygrad?tab=readme-ov-file#contributing", "context": "", "use_case": "", "switched_from": "", "story": "Five Years of Tinygrad"}, {"id": 743003568, "item_id": 46437723, "url": "https://news.ycombinator.com/item?id=46437723", "by": "Barrin92", "date": "2025-12-30", "stance": "neutral", "quote": "look at this: https://github.com/tinygrad/tinygrad/blob/master/tinygrad/codegen/simplify.py", "excerpt": "yes, it's really crazy, if people think you're exaggerating look at this: https://github.com/tinygrad/tinygrad/blob/master/tinygrad/codegen/simplify.py One of the cases why I think obsession with lines of code is one of the most counterproductive metrics, it always produces code like this.", "context": "", "use_case": "Deep learning framework", "switched_from": "", "story": "Five Years of Tinygrad"}, {"id": 742798656, "item_id": 46424916, "url": "https://news.ycombinator.com/item?id=46424916", "by": "piskov", "date": "2025-12-29", "stance": "recommend", "quote": "He was able to run nvidia gpu on mac via thunderbolt with tinygrad.", "excerpt": "He was able to run nvidia gpu on mac via thunderbolt with tinygrad. https://www.tomshardware.com/pc-components/gpus/tiny-corp-successfully-runs-an-nvidia-gpu-on-arm-macbook-through-usb4-using-an-external-gpu-docking-station \u2014 Check tinygrad\u2019s twitter account for specifics if you want to catch up on progress", "context": "", "use_case": "Machine learning framework", "switched_from": "", "story": "Five Years of Tinygrad"}, {"id": 742784176, "item_id": 46424011, "url": "https://news.ycombinator.com/item?id=46424011", "by": "jszymborski", "date": "2025-12-29", "stance": "neutral", "quote": "When we can reproduce a common set of papers on 1 NVIDIA GPU 2x faster than PyTorch.", "excerpt": "From [0]: \"When we can reproduce a common set of papers on 1 NVIDIA GPU 2x faster than PyTorch. We also want the speed to be good on the M1. ETA, Q2 next year.\" [0] https://tinygrad.org/#tinybox", "context": "", "use_case": "", "switched_from": "", "story": "Five Years of Tinygrad"}, {"id": 742338400, "item_id": 46396150, "url": "https://news.ycombinator.com/item?id=46396150", "by": "nemothekid", "date": "2025-12-26", "stance": "neutral", "quote": "it's part of his B2B company, tinygrad, that sells AMD systems and is VC funded", "excerpt": "Geohot isn't working by himself - it's part of his B2B company, tinygrad, that sells AMD systems and is VC funded. https://tinygrad.org/#tinybox You can see in his table he calls out his AMD system as having \"Good\" GPU support, vs. \"Great\" for nvidia. So, yes, I would argue he is doing the work to platform and organize people, on a professional level to sell AMD systems in a sustainable manner - everything you claim ", "context": "", "use_case": "ML framework", "switched_from": "", "story": "CUDA Tile Open Sourced"}, {"id": 742094976, "item_id": 46380936, "url": "https://news.ycombinator.com/item?id=46380936", "by": "refulgentis", "date": "2025-12-24", "stance": "warn", "quote": "I know George has pushed this idea for years now", "excerpt": "My deepest apologies, I can't parse this and I earnestly tried: 5 minutes of my own thinking, then 3 llms, then a 10 minute timer of my own thinking over the whole thing. My guess is you're trying to communicate \"tinygrad doesn't need gpu drivers\" which maybe is transmutated into \"tinygrad replaces CUDA\" and you think \"CUDA means other GPUs can't be used for LLMs, thus nvidia has a strangehold\" I know George has push", "context": "", "use_case": "", "switched_from": "", "story": "Nvidia buying AI chip startup Groq for about $20B in cash"}, {"id": 740275328, "item_id": 46267208, "url": "https://news.ycombinator.com/item?id=46267208", "by": "ekez", "date": "2025-12-14", "stance": "recommend", "quote": "tinygrad\u2019s small set of operations and laziness made it easy to implement.", "excerpt": "I implemented an algorithm for neural network verification called \u237a\u03b2-CROWN for a deep-learning library called tinygrad. https://github.com/0xekez/tinyLIRPA tinygrad\u2019s small set of operations and laziness made it easy to implement. Tho my overall sense is that neural network verification is currently more of a research interest than something practical.", "context": "", "use_case": "", "switched_from": "", "story": "Ask HN: What Are You Working On? (December 2025)"}, {"id": 740009376, "item_id": 46250586, "url": "https://news.ycombinator.com/item?id=46250586", "by": "throawayonthe", "date": "2025-12-12", "stance": "neutral", "quote": "", "excerpt": "https://social.treehouse.systems/@janne/115509948515319437 nvidia on a 2023 Mac Pro running linux :p", "context": "", "use_case": "", "switched_from": "", "story": "macOS 26.2 enables fast AI clusters with RDMA over Thunderbolt"}, {"id": 735803872, "item_id": 45987742, "url": "https://news.ycombinator.com/item?id=45987742", "by": "manjose2018", "date": "2025-11-19", "stance": "recommend", "quote": "https://tinygrad.org/ is the only viable alternative to CUDA that I have seen popup in the past few years.", "excerpt": "https://geohot.github.io//blog/jekyll/update/2025/03/08/AMD-YOLO.html https://tinygrad.org/ is the only viable alternative to CUDA that I have seen popup in the past few years.", "context": "", "use_case": "CUDA alternative for GPUs", "switched_from": "", "story": "Why CUDA translation wont unlock AMD"}]}