{"slug": "exo", "name": "exo", "category": "ai", "category_name": "AI tools and apps", "description": "", "recommenders": 8, "commenters": 14, "warners": 0, "score": 8.818, "rising": 1.36, "first": "2025-12-13", "latest": "2026-10-15", "why": "Doesn't do any actually clustering like https://github.com/exo-explore/exo does.", "why_is_quote": true, "has_verdict": false, "sentiment": {"recommend": 8, "neutral": 6, "compare": 1, "warn": 0}, "spark": [0, 2, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1], "listed": true, "loved": [], "caveat": "", "short_quote": "Doesn't do any actually clustering like https://github.com/exo-explore/exo does.", "kind": "", "icon": true, "domain": null, "github_repo": "github.com/exo-explore/exo", "official_url": "https://github.com/exo-explore/exo", "pricing": "", "aliases": ["exo", "Exo", "EXO"], "verdict": null, "praise": [{"label": "", "summary": "", "quotes": [{"id": 801757680, "item_id": 50109855, "url": "https://news.ycombinator.com/item?id=50109855", "by": "waz0wski", "date": "2026-10-15", "version": null, "stance": "recommend", "quote": "that's just one of the quick-and-easy setups if you've got a spare 40-60k", "excerpt": "Jeff Geerling blogged about his demo setup - https://www.jeffgeerling.com/blog/2025/15-tb-vram-on-mac-studio-rdma-over-thunderbolt-5/ Uses Exo - https://exolabs.net You can do it with other software, that's just one of the quick-and-easy setups if you've got a spare 40-60k for studios sitting around", "context": "", "use_case": "Clustering Macs for AI", "switched_from": "", "story": "Muse Spark Open"}, {"id": 795479728, "item_id": 49717483, "url": "https://news.ycombinator.com/item?id=49717483", "by": "compressedgas", "date": "2026-09-15", "version": null, "stance": "recommend", "quote": "Doesn't do any actually clustering like https://github.com/exo-explore/exo does.", "excerpt": "Doesn't do any actually clustering like https://github.com/exo-explore/exo does.", "context": "", "use_case": "local AI clustering", "switched_from": "", "story": "Nvidia PAIR turns your household's idle PCs into a local AI cluster"}, {"id": 792079040, "item_id": 49504940, "url": "https://news.ycombinator.com/item?id=49504940", "by": "latentsea", "date": "2026-08-31", "version": null, "stance": "recommend", "quote": "It looks pretty neat.", "excerpt": "There's a cool research project https://github.com/exoharness/exo that is designed specifically as a self-modifying harness, so architecturally it separates things in a way that makes it a lot harder for the agent to break itself when modifying itself. It looks pretty neat. Saw the author do an interview on a podcast explaining it.", "context": "", "use_case": "Self-modifying agent harness", "switched_from": "", "story": "Continuous Diffusion Language Models (CDLM's)"}, {"id": 785425792, "item_id": 49089112, "url": "https://news.ycombinator.com/item?id=49089112", "by": "0x20cowboy", "date": "2026-07-28", "version": null, "stance": "recommend", "quote": "Check out exo if you haven\u2019t already", "excerpt": "Check out exo if you haven\u2019t already https://github.com/exo-explore/exo", "context": "", "use_case": "distributed GPU/model inference across machines", "switched_from": "", "story": "Using a Gaming PC's RTX 5070 from a separate Linux workstation"}, {"id": 771936320, "item_id": 48246020, "url": "https://news.ycombinator.com/item?id=48246020", "by": "kappuchino", "date": "2026-05-23", "version": null, "stance": "recommend", "quote": "with exo I can run decent quantization", "excerpt": "I have two of the M3s due testing of models at work and with exo I can run decent quantization with 1 millon tokens for memory and derailment tests. Slow? Yes, but ... private. Unconditionally. And recently with https://github.com/antirez/ds4 one can use just one system to a very, very decent speed and ttft for chat inference. Again, private.", "context": "tests local models at work on Mac M3s", "use_case": "Distributed local LLM inference", "switched_from": "", "story": "Was my $48K GPU server worth it?"}]}], "caveats": [], "who": [{"by": "piyussh", "context": "SF hacker house", "url": "https://news.ycombinator.com/item?id=48473063", "date": "2026-06-10"}, {"by": "kappuchino", "context": "tests local models at work on Mac M3s", "url": "https://news.ycombinator.com/item?id=48246020", "date": "2026-05-23"}], "tags": [], "use_cases": ["clustering macs for ai", "distributed ai inference", "local ai clustering", "self-modifying agent harness", "distributed gpu/model inference across machines", "cluster gpus for local models"], "compared": [{"slug": "meshllm", "name": "MeshLLM", "co": 2, "switched_from": 0, "switched_to": 0}], "makers": [], "threads": [{"url": "https://news.ycombinator.com/item?id=48184402", "title": "Was my $48K GPU server worth it?", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=50107379", "title": "Muse Spark Open", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=50070593", "title": "A Dynamic and Distributed Virtual GPU Cloud Built on Idle Apple Silicon Devices", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49563556", "title": "Nvidia PAIR turns your household's idle PCs into a local AI cluster", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49502611", "title": "Continuous Diffusion Language Models (CDLM's)", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49125034", "title": "Getting 25 Gbps Thunderbolt Ethernet on My Mac Studio", "mentions": 1}], "sentiment_by_month": {"2025-11": {}, "2025-12": {"recommend": 2, "neutral": 2}, "2026-01": {}, "2026-02": {}, "2026-03": {"neutral": 1}, "2026-04": {"recommend": 1}, "2026-05": {"neutral": 1, "recommend": 1}, "2026-06": {"neutral": 1}, "2026-07": {"recommend": 1, "neutral": 1}, "2026-08": {"recommend": 1}, "2026-09": {"recommend": 1}, "2026-10": {"compare": 1, "recommend": 1}}, "versions": [], "version_unnamed": 0, "family_ref": null, "runs_on": null, "lines": [], "quotes": [{"id": 801757680, "item_id": 50109855, "url": "https://news.ycombinator.com/item?id=50109855", "by": "waz0wski", "date": "2026-10-15", "version": null, "stance": "recommend", "quote": "that's just one of the quick-and-easy setups if you've got a spare 40-60k", "excerpt": "Jeff Geerling blogged about his demo setup - https://www.jeffgeerling.com/blog/2025/15-tb-vram-on-mac-studio-rdma-over-thunderbolt-5/ Uses Exo - https://exolabs.net You can do it with other software, that's just one of the quick-and-easy setups if you've got a spare 40-60k for studios sitting around", "context": "", "use_case": "Clustering Macs for AI", "switched_from": "", "story": "Muse Spark Open"}, {"id": 801146336, "item_id": 50071646, "url": "https://news.ycombinator.com/item?id=50071646", "by": "gempir", "date": "2026-10-13", "version": null, "stance": "compare", "quote": "so this is basically Apple's version of exo ?", "excerpt": "Interesting, so this is basically Apple's version of exo ? https://github.com/exo-explore/exo", "context": "", "use_case": "distributed AI inference", "switched_from": "", "story": "A Dynamic and Distributed Virtual GPU Cloud Built on Idle Apple Silicon Devices"}, {"id": 795479728, "item_id": 49717483, "url": "https://news.ycombinator.com/item?id=49717483", "by": "compressedgas", "date": "2026-09-15", "version": null, "stance": "recommend", "quote": "Doesn't do any actually clustering like https://github.com/exo-explore/exo does.", "excerpt": "Doesn't do any actually clustering like https://github.com/exo-explore/exo does.", "context": "", "use_case": "local AI clustering", "switched_from": "", "story": "Nvidia PAIR turns your household's idle PCs into a local AI cluster"}, {"id": 792079040, "item_id": 49504940, "url": "https://news.ycombinator.com/item?id=49504940", "by": "latentsea", "date": "2026-08-31", "version": null, "stance": "recommend", "quote": "It looks pretty neat.", "excerpt": "There's a cool research project https://github.com/exoharness/exo that is designed specifically as a self-modifying harness, so architecturally it separates things in a way that makes it a lot harder for the agent to break itself when modifying itself. It looks pretty neat. Saw the author do an interview on a podcast explaining it.", "context": "", "use_case": "Self-modifying agent harness", "switched_from": "", "story": "Continuous Diffusion Language Models (CDLM's)"}, {"id": 786065392, "item_id": 49129087, "url": "https://news.ycombinator.com/item?id=49129087", "by": "geerlingguy", "date": "2026-07-31", "version": null, "stance": "neutral", "quote": "The 4 Mac Studio cluster I tested with Exo and Thunderbolt RDMA", "excerpt": "The Mac Studio in most of my videos is my own workstation, purchased a while back. The 4 Mac Studio cluster I tested with Exo and Thunderbolt RDMA was a loaner cluster from Apple. There's no universe in which I could justify buying that. I always mention any equipment that's provided / loaned / etc. and the nature of my relationships with vendors\u2014I try to also keep this document updated for transparency's sake: https", "context": "", "use_case": "", "switched_from": "", "story": "Getting 25 Gbps Thunderbolt Ethernet on My Mac Studio"}, {"id": 785425792, "item_id": 49089112, "url": "https://news.ycombinator.com/item?id=49089112", "by": "0x20cowboy", "date": "2026-07-28", "version": null, "stance": "recommend", "quote": "Check out exo if you haven\u2019t already", "excerpt": "Check out exo if you haven\u2019t already https://github.com/exo-explore/exo", "context": "", "use_case": "distributed GPU/model inference across machines", "switched_from": "", "story": "Using a Gaming PC's RTX 5070 from a separate Linux workstation"}, {"id": 775569009, "item_id": 48473063, "url": "https://news.ycombinator.com/item?id=48473063", "by": "piyussh", "date": "2026-06-10", "version": null, "stance": "neutral", "quote": "eventually connect all GPUs over Exo Labs to be able to run upwards of 400B parameter models locally", "excerpt": "At our SF hacker house, The Garden, we believe compute is the new commodity to own. This year, we\u2019ve pledged to buy $20K worth of compute so we can experiment with AI doing science, agent orchestration, and maybe even use it as edge compute to mimic scale for our apps and test how production-ready they are. We already have a DGX Spark and two RTX 5080s, and will eventually connect all GPUs over Exo Labs to be able to", "context": "SF hacker house", "use_case": "cluster GPUs for local models", "switched_from": "", "story": "We're spending 24 hours using local LLMs to search for the meaning of life"}, {"id": 771936320, "item_id": 48246020, "url": "https://news.ycombinator.com/item?id=48246020", "by": "kappuchino", "date": "2026-05-23", "version": null, "stance": "recommend", "quote": "with exo I can run decent quantization", "excerpt": "I have two of the M3s due testing of models at work and with exo I can run decent quantization with 1 millon tokens for memory and derailment tests. Slow? Yes, but ... private. Unconditionally. And recently with https://github.com/antirez/ds4 one can use just one system to a very, very decent speed and ttft for chat inference. Again, private.", "context": "tests local models at work on Mac M3s", "use_case": "Distributed local LLM inference", "switched_from": "", "story": "Was my $48K GPU server worth it?"}, {"id": 771679568, "item_id": 48229973, "url": "https://news.ycombinator.com/item?id=48229973", "by": "digitaltrees", "date": "2026-05-21", "version": null, "stance": "neutral", "quote": "I have three m3 512gb units and want a fourth to run an exo set up", "excerpt": "I have three m3 512gb units and want a fourth to run an exo set up. Like you, I am worried about scammers. Let\u2019s discuss if you still want to sell. https://calendly.com/ryanwmartin/open-office-hours", "context": "", "use_case": "", "switched_from": "", "story": "Was my $48K GPU server worth it?"}, {"id": 764621056, "item_id": 47788816, "url": "https://news.ycombinator.com/item?id=47788816", "by": "damezumari", "date": "2026-04-15", "version": null, "stance": "recommend", "quote": "https://github.com/exo-explore/exo", "excerpt": "https://github.com/exo-explore/exo", "context": "", "use_case": "pooling local machines for inference", "switched_from": "", "story": "Darkbloom \u2013 Private inference on idle Macs"}, {"id": 761295968, "item_id": 47580998, "url": "https://news.ycombinator.com/item?id=47580998", "by": "alexandercheema", "date": "2026-03-30", "version": null, "stance": "neutral", "quote": "there is already a big OSS project called exo", "excerpt": "lol bro there is already a big OSS project called exo: https://github.com/exo-explore/exo", "context": "", "use_case": "", "switched_from": "", "story": "Exo: Claude Code for your Inbox"}, {"id": 741629792, "item_id": 46351862, "url": "https://news.ycombinator.com/item?id=46351862", "by": "redrove", "date": "2025-12-21", "version": null, "stance": "neutral", "quote": "I keep hearing about it", "excerpt": "https://github.com/exo-explore/exo I keep hearing about it but unfortunately I myself only have one mac and nvidia GPUs and those can\u2019t cluster together :/", "context": "", "use_case": "", "switched_from": "", "story": "A guide to local coding models"}, {"id": 741472128, "item_id": 46342008, "url": "https://news.ycombinator.com/item?id=46342008", "by": "syntaxing", "date": "2025-12-20", "version": null, "stance": "recommend", "quote": "Theres some technical implementations that makes it more efficient like EXO", "excerpt": "Theres some technical implementations that makes it more efficient like EXO [1]. Jeff Geerling recently did a review on a 4 MAC Studio cluster with RDMA support and you can see that EXO has a noticeable advantage [2]. [1] https://github.com/exo-explore/exo [2] https://www.youtube.com/watch?v=x4_RsUxRjKU", "context": "", "use_case": "distributed LLM inference", "switched_from": "", "story": "Big GPUs don't need big PCs"}, {"id": 741150208, "item_id": 46321888, "url": "https://news.ycombinator.com/item?id=46321888", "by": "geerlingguy", "date": "2025-12-18", "version": null, "stance": "neutral", "quote": "Exo just outputs a time in ms, no other data for that", "excerpt": "See https://github.com/geerlingguy/beowulf-ai-cluster/issues/17 for more data \u2014 I didn't save all the prompt processing times (Exo just outputs a time in ms, no other data for that), but will try to have another pass. Maybe also convince the Exo team to add a proper benchmarking capability ala `llama-bench` :)", "context": "", "use_case": "", "switched_from": "", "story": "1.5 TB of VRAM on Mac Studio \u2013 RDMA over Thunderbolt 5"}, {"id": 740069696, "item_id": 46254356, "url": "https://news.ycombinator.com/item?id=46254356", "by": "aimanbenbaha", "date": "2025-12-13", "version": null, "stance": "recommend", "quote": "Exo-Labs is an open source project that allows this too", "excerpt": "Exo-Labs is an open source project that allows this too, pipeline parallelism I mean not the latter, and it's device agnostic meaning you can daisy-chain anything you have that has memory and the implementation will intelligently shard model layers across them, though its slow but scales linearly with concurrent requests. Exo-Labs: https://github.com/exo-explore/exo", "context": "", "use_case": "", "switched_from": "", "story": "macOS 26.2 enables fast AI clusters with RDMA over Thunderbolt"}]}