{"slug": "lm-studio", "name": "LM Studio", "category": "ai", "category_name": "AI models and assistants", "description": "Desktop app for downloading and running local LLMs, with a built-in OpenAI-compatible server.", "recommenders": 24, "commenters": 50, "warners": 2, "score": 25.845, "rising": 2.76, "first": "2025-11-01", "latest": "2026-10-04", "why": "Liked as a low-friction desktop app for downloading and running local models, with a built-in local server that plugs into coding agents like OpenCode, and fast MLX support on Apple Silicon Macs.", "why_is_quote": false, "has_verdict": true, "sentiment": {"recommend": 23, "neutral": 23, "compare": 5, "warn": 2}, "spark": [0, 1, 1, 0, 2, 0, 0, 4, 8, 4, 3, 0], "listed": true, "loved": ["Local API server", "Easy setup"], "caveat": "Closed source", "short_quote": "I strongly recommend trying LM Studio - it's the lowest friction way to try out models", "icon": true, "domain": "lmstudio.ai", "github_repo": null, "official_url": "https://lmstudio.ai", "pricing": "free", "aliases": ["LM Studio"], "verdict": [{"text": "Liked as a low-friction desktop app for downloading and running local models", "cites": [{"id": 776911824, "item_id": 48556989, "url": "https://news.ycombinator.com/item?id=48556989", "by": "simonw", "date": "2026-06-16", "stance": "recommend", "quote": "I strongly recommend trying LM Studio - it's the lowest friction way to try out models", "excerpt": "I strongly recommend trying LM Studio - it's the lowest friction way to try out models, you can browse https://lmstudio.ai/models and click \"Get\" and then \"Run in LM Studio\" to download and run a model. With 96GB I'd start with the Gemma 4 and Qwen 3.6 models. Any of those should work fine.", "context": "", "use_case": "running local models", "switched_from": "", "story": "Running local models is good now"}, {"id": 787288384, "item_id": 49205524, "url": "https://news.ycombinator.com/item?id=49205524", "by": "alasr", "date": "2026-08-06", "stance": "recommend", "quote": "\"LM Studio\"[2] is not bad to start with as you'll be able to download, start/stop/manage and chat with your LLM model all in one place", "excerpt": "- See \"Llama.cpp vs Ollama\"[0] and \"Llama.cpp vs LM Studio\"[1] for a high-level features comparison and find out which one suite you best (check out the \"Target Users\" category[0][1], at least). - If you're still not sure, \"LM Studio\"[2] is not bad to start with as you'll be able to download, start/stop/manage and chat with your LLM model all in one place: a single desktop app. Also, once installed, enable \"Developer", "context": "", "use_case": "running local LLMs", "switched_from": "", "story": "Qwen3.8-Max: A New Bar for Coding and Cowork"}, {"id": 783782145, "item_id": 48986384, "url": "https://news.ycombinator.com/item?id=48986384", "by": "dofm", "date": "2026-07-20", "stance": "recommend", "quote": "LM Studio (nicer app on the Mac)", "excerpt": "On an M1 Max I have been using either Unsloth Studio (which is basically a web app) or LM Studio (nicer app on the Mac). You can use the Google AI Edge Gallery to play with the smaller Gemma models (but at the moment the QAT variants don't seem to be there unless I am missing something). I think it's likely the 26B QAT model won't fit in your machine \u2014 you may be able to fit one of the UD_Q3 or UD_Q2 variants but whe", "context": "M1 Max user", "use_case": "running local models", "switched_from": "", "story": "Nativ: Run frontier open models locally on your Mac"}]}, {"text": ", with a built-in local server that plugs into coding agents like OpenCode", "cites": [{"id": 776215408, "item_id": 48513463, "url": "https://news.ycombinator.com/item?id=48513463", "by": "hmontazeri", "date": "2026-06-12", "stance": "recommend", "quote": "I use LM Studio with the local server it ships and connect it to opencode. Takes 2 min to setup", "excerpt": "I use LM Studio with the local server it ships and connect it to opencode. Takes 2 min to setup", "context": "", "use_case": "local model server", "switched_from": "", "story": "How to Setup a Local Coding Agent on macOS"}, {"id": 757880080, "item_id": 47367505, "url": "https://news.ycombinator.com/item?id=47367505", "by": "brcmthrowaway", "date": "2026-03-13", "stance": "recommend", "quote": "LM Studio has a server that serves models over HTTP (localhost).", "excerpt": "Basically LM Studio has a server that serves models over HTTP (localhost). Configure/enable the server and connect OpenCode to it. Try this article https://advanced-stack.com/fields-notes/qwen35-opencode-lm-studio-agentic-coding-on-m1.html I'm looking for an alternative to OpenCode though, I can barely see the UI.", "context": "", "use_case": "local model server", "switched_from": "", "story": "Can I Run AI locally?"}, {"id": 783094080, "item_id": 48943380, "url": "https://news.ycombinator.com/item?id=48943380", "by": "IronWolve", "date": "2026-07-16", "stance": "recommend", "quote": "I love lm-studio, so cant wait to see how this goes. For local I normally use opencode + lmstudio.", "excerpt": "It was fun to try out which local models made better agents. Few thoughts. * locked to a single dir, so no system wide access. * no local web search, can be fixed ddg or local mcp. * no ssh, I want to have it ssh into my server and do the work. * doesnt show the model being loaded, needs a bar/% counter. * Can you drag/drop documents in the work dirs, or only + add them? I love lm-studio, so cant wait to see how this", "context": "", "use_case": "running local models", "switched_from": "", "story": "LM Studio Bionic: the AI agent for open models"}]}, {"text": ", and fast MLX support on Apple Silicon Macs.", "cites": [{"id": 796356992, "item_id": 49772312, "url": "https://news.ycombinator.com/item?id=49772312", "by": "aetherspawn", "date": "2026-09-20", "stance": "recommend", "quote": "I use LM Studio with MLX support, which is specifically faster on M series Macs.", "excerpt": "Yes, I use LM Studio with MLX support, which is specifically faster on M series Macs. I am not sure if llama is the same, but I guess what I\u2019m saying is if you want the performance to be good on M series you have to use models packaged in the right format.", "context": "M series Mac user", "use_case": "running local models on Mac", "switched_from": "", "story": "Bonsai 2 27B: Near-Lossless Compression in a 9x Smaller Footprint"}, {"id": 783755361, "item_id": 48984710, "url": "https://news.ycombinator.com/item?id=48984710", "by": "erikkahler", "date": "2026-07-20", "stance": "recommend", "quote": "the excellent LM Studio", "excerpt": "To clarify, this MIT-licensed app is from the very same dev, 'Prince Canuma', who maintains the popular MLX-VLM library (https://github.com/Blaizzy/mlx-vlm). MLX-VLM is a long-time dependency of the excellent LM Studio and others because it can provide faster inference on Apple devices than llama.cpp. Historically, MLX is a smaller community than CUDA, but has some of the fastest updates upon the release of new model", "context": "", "use_case": "local LLM runner", "switched_from": "", "story": "Nativ: Run frontier open models locally on your Mac"}]}, {"text": " The main complaints are that it is closed source, with slow bug fixes and unclear telemetry", "cites": [{"id": 786509616, "item_id": 49156851, "url": "https://news.ycombinator.com/item?id=49156851", "by": "imrehg", "date": "2026-08-03", "stance": "warn", "quote": "Case to point is LM Studio. It's closed source, has bugs (duh!), and there's at least a GitHub issue tracker to report the bugs -- but then by and large nothing happens", "excerpt": "I have sent countless small bug fixes even before, to tools that I use that I could dig into when something was off... No big heavy lifting, but plenty of \"tinker on your tools\" stuff. And I don't think what I did was that special, so maybe the experiences are different? The current LLM-driven stuff seems to break down the expectations, and now there are a lot of places which just ignore anything I send in (the same ", "context": "contributes small bug fixes to tools he uses", "use_case": "running local LLMs", "switched_from": "", "story": "Devtools must be open source"}, {"id": 787906129, "item_id": 49244133, "url": "https://news.ycombinator.com/item?id=49244133", "by": "DanielHB", "date": "2026-08-10", "stance": "recommend", "quote": "I eventually switched to LM studio and the same model runs much better, like 70tk/s.", "excerpt": "I have tried running llamma.cpp on my PC and I found it hard getting it to run at decent speed. On Qwen 3.5-9b I get at most 10tk/s. I eventually switched to LM studio and the same model runs much better, like 70tk/s. Not sure if it was because I was running llama.cpp inside podman or badly tuned LLM arguments. But LM studio is unfortunately much more practical. Although I agree with you. I do not really know what ki", "context": "", "use_case": "running local LLMs", "switched_from": "llama.cpp", "story": "DeepSeek costs OpenCode Go user $1.14/day; dual DGX breaks even in 24 years"}, {"id": 771593889, "item_id": 48224618, "url": "https://news.ycombinator.com/item?id=48224618", "by": "satvikpendem", "date": "2026-05-21", "stance": "compare", "quote": "open source alternative to the more widely known LM Studio", "excerpt": "Unsloth Studio [0] is what I recommend these days, open source alternative to the more widely known LM Studio, and also built by the people who make good quantizations of released models. With MTP support not merged in you should get 2x token generation speed with no accuracy difference. They also have MLX quants if you scroll down a bit, which is a format specifically for macOS' Metal GPU acceleration but that's not", "context": "", "use_case": "running local models", "switched_from": "", "story": "Indexing a year of video locally on a 2021 MacBook with Gemma4-31B (50GB swap)"}]}, {"text": ", plus hardware quirks such as broken ROCm support on AMD integrated GPUs and out-of-memory crashes.", "cites": [{"id": 788296289, "item_id": 49268518, "url": "https://news.ycombinator.com/item?id=49268518", "by": "imrehg", "date": "2026-08-11", "stance": "warn", "quote": "Their ROCm runtime does not enable the the AMD GPU inside the Framework", "excerpt": "llama.cpp works pretty well for me on the Framework 13 laptop, but the current era of \"move fast, break things, rarely fix\" (sorry, that's how it feels), bites here quite a bit. Two examples: - https://github.com/ggml-org/llama.cpp/pull/25863 Someone's few lines change broke the native (ROCm) support for the AMD GPU inside Framework (and other integrated systems), and any rollback or proper fix is pending for almost ", "context": "", "use_case": "running GGUF models", "switched_from": "", "story": "llama.cpp"}, {"id": 747300482, "item_id": 46706280, "url": "https://news.ycombinator.com/item?id=46706280", "by": "river_otter", "date": "2026-01-21", "stance": "neutral", "quote": "I'm documenting my experience using aoe + OpenCode + LM Studio + GLM-4.7 Flash + Mac Mini M4Pro 64GB Mem", "excerpt": "Excited to finally be able to give this a try today. I'm documenting my experience using aoe + OpenCode + LM Studio + GLM-4.7 Flash + Mac Mini M4Pro 64GB Mem on this thread if anyone wants to follow along and or give me advice about how badly I'm messing up the settings https://x.com/natebrake/status/2013978241573204246 Thus far, the 6-bit quant MLX weights were too much and crashed LMS with OOM", "context": "", "use_case": "running local models", "switched_from": "", "story": "GLM-4.7-Flash"}]}], "praise": [{"label": "Local API server", "summary": "Its bundled localhost server, with OpenAI- and Anthropic-compatible endpoints, connects easily to agents and editors.", "quotes": [{"id": 776215408, "item_id": 48513463, "url": "https://news.ycombinator.com/item?id=48513463", "by": "hmontazeri", "date": "2026-06-12", "stance": "recommend", "quote": "I use LM Studio with the local server it ships and connect it to opencode. Takes 2 min to setup", "excerpt": "I use LM Studio with the local server it ships and connect it to opencode. Takes 2 min to setup", "context": "", "use_case": "local model server", "switched_from": "", "story": "How to Setup a Local Coding Agent on macOS"}, {"id": 757880080, "item_id": 47367505, "url": "https://news.ycombinator.com/item?id=47367505", "by": "brcmthrowaway", "date": "2026-03-13", "stance": "recommend", "quote": "LM Studio has a server that serves models over HTTP (localhost).", "excerpt": "Basically LM Studio has a server that serves models over HTTP (localhost). Configure/enable the server and connect OpenCode to it. Try this article https://advanced-stack.com/fields-notes/qwen35-opencode-lm-studio-agentic-coding-on-m1.html I'm looking for an alternative to OpenCode though, I can barely see the UI.", "context": "", "use_case": "local model server", "switched_from": "", "story": "Can I Run AI locally?"}, {"id": 778893924, "item_id": 48680870, "url": "https://news.ycombinator.com/item?id=48680870", "by": "kylenessen", "date": "2026-06-25", "stance": "recommend", "quote": "I use LMStudio which supports OpenAI and Anthropic compatible api endpoints", "excerpt": "I really wanted to like this, but unfortunately couldn't see how it improves my experience over Obsidian or VS Code. The fact that I have to juggle between OpenKnowledge and Codex to engage the AI, while also accepting a barebones Obsidian, is a real bummer. From what I can tell, you are saving me a few key strokes with moving prompts around. What I really want is the AI to live IN the app, like VS Code, and then mov", "context": "", "use_case": "running local models", "switched_from": "", "story": "Show HN: OpenKnowledge \u2013 open-source alternative to Obsidian/Notion"}]}, {"label": "Easy setup", "summary": "Commenters call it the lowest-friction way to download, manage and chat with models in one app.", "quotes": [{"id": 776911824, "item_id": 48556989, "url": "https://news.ycombinator.com/item?id=48556989", "by": "simonw", "date": "2026-06-16", "stance": "recommend", "quote": "I strongly recommend trying LM Studio - it's the lowest friction way to try out models", "excerpt": "I strongly recommend trying LM Studio - it's the lowest friction way to try out models, you can browse https://lmstudio.ai/models and click \"Get\" and then \"Run in LM Studio\" to download and run a model. With 96GB I'd start with the Gemma 4 and Qwen 3.6 models. Any of those should work fine.", "context": "", "use_case": "running local models", "switched_from": "", "story": "Running local models is good now"}, {"id": 787288384, "item_id": 49205524, "url": "https://news.ycombinator.com/item?id=49205524", "by": "alasr", "date": "2026-08-06", "stance": "recommend", "quote": "\"LM Studio\"[2] is not bad to start with as you'll be able to download, start/stop/manage and chat with your LLM model all in one place", "excerpt": "- See \"Llama.cpp vs Ollama\"[0] and \"Llama.cpp vs LM Studio\"[1] for a high-level features comparison and find out which one suite you best (check out the \"Target Users\" category[0][1], at least). - If you're still not sure, \"LM Studio\"[2] is not bad to start with as you'll be able to download, start/stop/manage and chat with your LLM model all in one place: a single desktop app. Also, once installed, enable \"Developer", "context": "", "use_case": "running local LLMs", "switched_from": "", "story": "Qwen3.8-Max: A New Bar for Coding and Cowork"}, {"id": 783782145, "item_id": 48986384, "url": "https://news.ycombinator.com/item?id=48986384", "by": "dofm", "date": "2026-07-20", "stance": "recommend", "quote": "LM Studio (nicer app on the Mac)", "excerpt": "On an M1 Max I have been using either Unsloth Studio (which is basically a web app) or LM Studio (nicer app on the Mac). You can use the Google AI Edge Gallery to play with the smaller Gemma models (but at the moment the QAT variants don't seem to be there unless I am missing something). I think it's likely the 26B QAT model won't fit in your machine \u2014 you may be able to fit one of the UD_Q3 or UD_Q2 variants but whe", "context": "M1 Max user", "use_case": "running local models", "switched_from": "", "story": "Nativ: Run frontier open models locally on your Mac"}]}, {"label": "Mac MLX speed", "summary": "MLX support makes it faster on M-series Macs, and MLX builds of new models appear quickly.", "quotes": [{"id": 796356992, "item_id": 49772312, "url": "https://news.ycombinator.com/item?id=49772312", "by": "aetherspawn", "date": "2026-09-20", "stance": "recommend", "quote": "I use LM Studio with MLX support, which is specifically faster on M series Macs.", "excerpt": "Yes, I use LM Studio with MLX support, which is specifically faster on M series Macs. I am not sure if llama is the same, but I guess what I\u2019m saying is if you want the performance to be good on M series you have to use models packaged in the right format.", "context": "M series Mac user", "use_case": "running local models on Mac", "switched_from": "", "story": "Bonsai 2 27B: Near-Lossless Compression in a 9x Smaller Footprint"}, {"id": 783755361, "item_id": 48984710, "url": "https://news.ycombinator.com/item?id=48984710", "by": "erikkahler", "date": "2026-07-20", "stance": "recommend", "quote": "the excellent LM Studio", "excerpt": "To clarify, this MIT-licensed app is from the very same dev, 'Prince Canuma', who maintains the popular MLX-VLM library (https://github.com/Blaizzy/mlx-vlm). MLX-VLM is a long-time dependency of the excellent LM Studio and others because it can provide faster inference on Apple devices than llama.cpp. Historically, MLX is a smaller community than CUDA, but has some of the fastest updates upon the release of new model", "context": "", "use_case": "local LLM runner", "switched_from": "", "story": "Nativ: Run frontier open models locally on your Mac"}, {"id": 781698784, "item_id": 48856174, "url": "https://news.ycombinator.com/item?id=48856174", "by": "kamranjon", "date": "2026-07-09", "stance": "recommend", "quote": "I use LMStudio pretty regularly", "excerpt": "I dunno I use LMStudio pretty regularly and the MLX folks and the community usually have MLX versions of new model releases up within a day or two.", "context": "", "use_case": "running local models on Mac", "switched_from": "", "story": "Apple Silicon Exec Explains Mac Mini AI Demand and On-Device Future"}]}, {"label": "Daily driver", "summary": "Several commenters use it regularly for local models in coding and general questions.", "quotes": [{"id": 792106146, "item_id": 49506634, "url": "https://news.ycombinator.com/item?id=49506634", "by": "Schlagbohrer", "date": "2026-08-31", "stance": "recommend", "quote": "with Qwen3.8-27B in back in LM Studio, to be amazing and i use it daily", "excerpt": "I have found the Pi agent harness, running in a docker container, with Qwen3.8-27B in back in LM Studio, to be amazing and i use it daily. Just last weekend I used it to do PDF editing which previously would have required me to download really shady software online to do. I have never tried OpenClaw so maybe I am missing out but considering how much my agent can modify the Pi environment to work better, I am happy wi", "context": "", "use_case": "local model serving", "switched_from": "", "story": "OpenClaw 2.0, Accidentally"}, {"id": 785593648, "item_id": 49099603, "url": "https://news.ycombinator.com/item?id=49099603", "by": "ktosobcy", "date": "2026-07-29", "stance": "recommend", "quote": "Recently I started playing with LM Studio and local models", "excerpt": "Recently I started playing with LM Studio and local models and found out that `gemma-4-26b-a4b` is suprisingly capable. I don't need elaborate akin to \"create complete app to do X\" or \"refactor the whole codebase of bazzilion of LOC\" but rather \"how to go about doing thing X\" or for language study (explaining nuances of phrasal verbs or subtelties of vocabulary in other languagues) and darn -- the results are rather ", "context": "", "use_case": "running local models", "switched_from": "", "story": "Self-hosting Kimi K3: 20% more hardware cost, 20% better task resolution"}, {"id": 748963392, "item_id": 46810212, "url": "https://news.ycombinator.com/item?id=46810212", "by": "TomMasz", "date": "2026-01-29", "stance": "recommend", "quote": "I've been using LM Studio for a while, this is a nice update.", "excerpt": "I've been using LM Studio for a while, this is a nice update. For what I need, running a local model is more than adequate. As long as you have sufficient RAM, of course.", "context": "", "use_case": "running local models", "switched_from": "", "story": "LM Studio 0.4.0"}]}, {"label": "Faster than llama.cpp", "summary": "One commenter got about 70 tok/s after switching from a self-run llama.cpp setup that managed about 10 tok/s.", "quotes": [{"id": 787906129, "item_id": 49244133, "url": "https://news.ycombinator.com/item?id=49244133", "by": "DanielHB", "date": "2026-08-10", "stance": "recommend", "quote": "I eventually switched to LM studio and the same model runs much better, like 70tk/s.", "excerpt": "I have tried running llamma.cpp on my PC and I found it hard getting it to run at decent speed. On Qwen 3.5-9b I get at most 10tk/s. I eventually switched to LM studio and the same model runs much better, like 70tk/s. Not sure if it was because I was running llama.cpp inside podman or badly tuned LLM arguments. But LM studio is unfortunately much more practical. Although I agree with you. I do not really know what ki", "context": "", "use_case": "running local LLMs", "switched_from": "llama.cpp", "story": "DeepSeek costs OpenCode Go user $1.14/day; dual DGX breaks even in 24 years"}]}, {"label": "Structured outputs", "summary": "It supports structured JSON-schema outputs, as llama.cpp does.", "quotes": [{"id": 777788913, "item_id": 48611807, "url": "https://news.ycombinator.com/item?id=48611807", "by": "anonym29", "date": "2026-06-20", "stance": "recommend", "quote": "llama.cpp supports structured outputs, as do some more consumer-oriented tools that wrap like LM Studio.", "excerpt": "In that case, you don't want to be using Claude Code, which is more of a consumer product; you instead want to control the inference stack yourself. What you are looking for is structured output (you give the inference engine a JSON schema you define that the response must conform to) + a JSON schema validator that parses the output, checks if the response is valid JSON. If it is, you're good to go, if not, run the i", "context": "", "use_case": "structured output inference", "switched_from": "", "story": "Ask HN: What technique do you use to make Claude Code deterministic?"}]}], "caveats": [{"label": "Closed source", "summary": "It is closed source, reported bugs often go unfixed, and its telemetry is unclear.", "quotes": [{"id": 786509616, "item_id": 49156851, "url": "https://news.ycombinator.com/item?id=49156851", "by": "imrehg", "date": "2026-08-03", "stance": "warn", "quote": "Case to point is LM Studio. It's closed source, has bugs (duh!), and there's at least a GitHub issue tracker to report the bugs -- but then by and large nothing happens", "excerpt": "I have sent countless small bug fixes even before, to tools that I use that I could dig into when something was off... No big heavy lifting, but plenty of \"tinker on your tools\" stuff. And I don't think what I did was that special, so maybe the experiences are different? The current LLM-driven stuff seems to break down the expectations, and now there are a lot of places which just ignore anything I send in (the same ", "context": "contributes small bug fixes to tools he uses", "use_case": "running local LLMs", "switched_from": "", "story": "Devtools must be open source"}, {"id": 787906129, "item_id": 49244133, "url": "https://news.ycombinator.com/item?id=49244133", "by": "DanielHB", "date": "2026-08-10", "stance": "recommend", "quote": "I eventually switched to LM studio and the same model runs much better, like 70tk/s.", "excerpt": "I have tried running llamma.cpp on my PC and I found it hard getting it to run at decent speed. On Qwen 3.5-9b I get at most 10tk/s. I eventually switched to LM studio and the same model runs much better, like 70tk/s. Not sure if it was because I was running llama.cpp inside podman or badly tuned LLM arguments. But LM studio is unfortunately much more practical. Although I agree with you. I do not really know what ki", "context": "", "use_case": "running local LLMs", "switched_from": "llama.cpp", "story": "DeepSeek costs OpenCode Go user $1.14/day; dual DGX breaks even in 24 years"}, {"id": 771593889, "item_id": 48224618, "url": "https://news.ycombinator.com/item?id=48224618", "by": "satvikpendem", "date": "2026-05-21", "stance": "compare", "quote": "open source alternative to the more widely known LM Studio", "excerpt": "Unsloth Studio [0] is what I recommend these days, open source alternative to the more widely known LM Studio, and also built by the people who make good quantizations of released models. With MTP support not merged in you should get 2x token generation speed with no accuracy difference. They also have MLX quants if you scroll down a bit, which is a format specifically for macOS' Metal GPU acceleration but that's not", "context": "", "use_case": "running local models", "switched_from": "", "story": "Indexing a year of video locally on a 2021 MacBook with Gemma4-31B (50GB swap)"}]}, {"label": "Hardware quirks", "summary": "Users report that the ROCm runtime does not use the AMD GPU on Framework machines, along with out-of-memory crashes and slowdowns when a model is split across GPUs.", "quotes": [{"id": 788296289, "item_id": 49268518, "url": "https://news.ycombinator.com/item?id=49268518", "by": "imrehg", "date": "2026-08-11", "stance": "warn", "quote": "Their ROCm runtime does not enable the the AMD GPU inside the Framework", "excerpt": "llama.cpp works pretty well for me on the Framework 13 laptop, but the current era of \"move fast, break things, rarely fix\" (sorry, that's how it feels), bites here quite a bit. Two examples: - https://github.com/ggml-org/llama.cpp/pull/25863 Someone's few lines change broke the native (ROCm) support for the AMD GPU inside Framework (and other integrated systems), and any rollback or proper fix is pending for almost ", "context": "", "use_case": "running GGUF models", "switched_from": "", "story": "llama.cpp"}, {"id": 747300482, "item_id": 46706280, "url": "https://news.ycombinator.com/item?id=46706280", "by": "river_otter", "date": "2026-01-21", "stance": "neutral", "quote": "I'm documenting my experience using aoe + OpenCode + LM Studio + GLM-4.7 Flash + Mac Mini M4Pro 64GB Mem", "excerpt": "Excited to finally be able to give this a try today. I'm documenting my experience using aoe + OpenCode + LM Studio + GLM-4.7 Flash + Mac Mini M4Pro 64GB Mem on this thread if anyone wants to follow along and or give me advice about how badly I'm messing up the settings https://x.com/natebrake/status/2013978241573204246 Thus far, the 6-bit quant MLX weights were too much and crashed LMS with OOM", "context": "", "use_case": "running local models", "switched_from": "", "story": "GLM-4.7-Flash"}, {"id": 796018064, "item_id": 49751129, "url": "https://news.ycombinator.com/item?id=49751129", "by": "Pragmata", "date": "2026-09-18", "stance": "neutral", "quote": "offload some into system ram when using LM studio", "excerpt": "I got a RTX5080 as primary and a RTX3050 (8gb) as secondary. I always max out on context size so that means i need to split the model between both gpus + offload some into system ram when using LM studio. I'm guessing that is what accounts for most of the slowdown when i use that. With this recipe i don't have to worry about it. It's pretty fast, and since its all running in the 5080 with max context it's stable (tho", "context": "", "use_case": "running local LLMs", "switched_from": "", "story": "Run QWEN3.8 27B on 16gb Nvidia GPUs"}]}, {"label": "Missing features", "summary": "Users note that image-related parameters are not implemented and that there is no Android companion app.", "quotes": [{"id": 788820531, "item_id": 49301283, "url": "https://news.ycombinator.com/item?id=49301283", "by": "jwr", "date": "2026-08-14", "stance": "warn", "quote": "I got tired of LM Studio not implementing the image-related parameters", "excerpt": "I have an M4 Max (unfortunately 64GB). I have been running the Qwen 35B A3B one for a while now, after testing and benchmarking a number of models. That one was consistently the best in class for tasks like despamming, E-mail classification, OCR and dictation post-processing. It was also really fast (90 tokens/s). I'm benchmarking the 3.8 model now, it seems it is better (near-perfect score on my E-mail spam filterin", "context": "", "use_case": "running local models", "switched_from": "", "story": "Qwen3.8-27B"}, {"id": 783137232, "item_id": 48946077, "url": "https://news.ycombinator.com/item?id=48946077", "by": "hypercube33", "date": "2026-07-17", "stance": "recommend", "quote": "I love LM studio and this is super cool", "excerpt": "I love LM studio and this is super cool and I plan on trying it out tonight. I'm absolutely dying for an android link app and I'm jealous of iPhone users", "context": "", "use_case": "running local models", "switched_from": "", "story": "LM Studio Bionic: the AI agent for open models"}]}, {"label": "Token speed", "summary": "One user is considering switching away from it to get better token speed.", "quotes": [{"id": 779825617, "item_id": 48739101, "url": "https://news.ycombinator.com/item?id=48739101", "by": "razster", "date": "2026-06-30", "stance": "neutral", "quote": "I may have to switch away from LM-studio if I can improve token speed.", "excerpt": "Thanks for sharing that. I have the same card but 96gb ram. I use PI.dev to connect to LM-Studio. I may have to switch away from LM-studio if I can improve token speed. I think I range from 32-40t/s. qwen3.6-35b-a3b-genesis-v2-apex-mtp.", "context": "has 12GB 4070 with 96GB RAM", "use_case": "running local models", "switched_from": "", "story": "Qwen 3.6 27B is the sweet spot for local development"}]}], "who": [{"by": "aetherspawn", "context": "M series Mac user", "url": "https://news.ycombinator.com/item?id=49772312", "date": "2026-09-20"}, {"by": "selicos", "context": "infra engineer, local LLM tinkerer", "url": "https://news.ycombinator.com/item?id=49744479", "date": "2026-09-17"}, {"by": "wolvoleo", "context": "former Mac user, now Linux", "url": "https://news.ycombinator.com/item?id=49485770", "date": "2026-08-29"}, {"by": "dofm", "context": "M1 Max user", "url": "https://news.ycombinator.com/item?id=48986384", "date": "2026-07-20"}, {"by": "kamranjon", "context": "runs fully local", "url": "https://news.ycombinator.com/item?id=48764703", "date": "2026-07-02"}, {"by": "razster", "context": "has 12GB 4070 with 96GB RAM", "url": "https://news.ycombinator.com/item?id=48739101", "date": "2026-06-30"}, {"by": "Saris", "context": "12GB 4070, 32GB RAM", "url": "https://news.ycombinator.com/item?id=48732472", "date": "2026-06-30"}, {"by": "rvba", "context": "runs local models on Windows 11 with a 5090", "url": "https://news.ycombinator.com/item?id=48355022", "date": "2026-06-01"}], "tags": ["for Mac users", "for beginners", "for local coding agents", "for Linux users", "for Windows users"], "use_cases": ["running local models", "running local llms", "running local models on mac", "local model server", "local model serving", "running local llms on macos"], "compared": [{"slug": "llama-cpp", "name": "llama.cpp", "co": 17, "switched_from": 1, "switched_to": 1}, {"slug": "ollama", "name": "Ollama", "co": 12, "switched_from": 1, "switched_to": 0}, {"slug": "qwen-3-6", "name": "Qwen 3.6", "co": 8, "switched_from": 0, "switched_to": 0}, {"slug": "opencode", "name": "OpenCode", "co": 7, "switched_from": 0, "switched_to": 0}, {"slug": "gemma", "name": "Gemma", "co": 6, "switched_from": 0, "switched_to": 0}, {"slug": "claude-code", "name": "Claude Code", "co": 4, "switched_from": 0, "switched_to": 0}, {"slug": "gpt-oss", "name": "GPT-OSS", "co": 4, "switched_from": 0, "switched_to": 0}, {"slug": "qwen-3-5", "name": "Qwen 3.5", "co": 4, "switched_from": 0, "switched_to": 0}], "makers": [], "threads": [{"url": "https://news.ycombinator.com/item?id=48939662", "title": "LM Studio Bionic: the AI agent for open models", "mentions": 4}, {"url": "https://news.ycombinator.com/item?id=48982681", "title": "Nativ: Run frontier open models locally on your Mac", "mentions": 3}, {"url": "https://news.ycombinator.com/item?id=48721903", "title": "Qwen 3.6 27B is the sweet spot for local development", "mentions": 3}, {"url": "https://news.ycombinator.com/item?id=49479951", "title": "Run Qwen3.8 27B locally: real numbers from my Mac Studio", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=48542100", "title": "Ask HN: Has anyone replaced Claude/GPT with a local model for daily coding?", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=46799477", "title": "LM Studio 0.4.0", "mentions": 2}], "sentiment_by_month": {"2025-11": {"neutral": 1}, "2025-12": {"recommend": 1}, "2026-01": {"neutral": 4, "compare": 1, "recommend": 1}, "2026-02": {"neutral": 1}, "2026-03": {"recommend": 2}, "2026-04": {}, "2026-05": {"neutral": 1, "compare": 1}, "2026-06": {"neutral": 8, "compare": 1, "recommend": 4}, "2026-07": {"recommend": 8, "neutral": 2, "compare": 1}, "2026-08": {"warn": 2, "recommend": 4, "neutral": 3, "compare": 1}, "2026-09": {"neutral": 2, "recommend": 3}, "2026-10": {"neutral": 1}}, "quotes": [{"id": 799299490, "item_id": 49956218, "url": "https://news.ycombinator.com/item?id=49956218", "by": "nottorp", "date": "2026-10-04", "stance": "neutral", "quote": "I used both lm studio and whatever the name of their new coding assistant", "excerpt": "At Q what? That's what I'm mostly asking about. My little test was \"generate me a single page tic tac toe game in plain javascript. computer always play O. add unbeatable minmax. have the board, a status line and a new game button'. I used both lm studio and whatever the name of their new coding assistant that supercedes lm studio is. Qwen 3.5 Q4 went into some kind of loop where it fixed whatever was broken on the p", "context": "", "use_case": "running local models", "switched_from": "", "story": "Run Qwen 3.8 Flash Next (125B) on consumer hardware (RTX 4090) at 100T/s"}, {"id": 797670369, "item_id": 49854398, "url": "https://news.ycombinator.com/item?id=49854398", "by": "dorianniemiec", "date": "2026-09-26", "stance": "recommend", "quote": "locally running models (at most 27B or 35B A3B or similar) on LM Studio", "excerpt": "Free models on OpenCode, or just locally running models (at most 27B or 35B A3B or similar) on LM Studio... :')", "context": "", "use_case": "running local models", "switched_from": "", "story": "Ask HN: Which model do you use for work?"}, {"id": 796356992, "item_id": 49772312, "url": "https://news.ycombinator.com/item?id=49772312", "by": "aetherspawn", "date": "2026-09-20", "stance": "recommend", "quote": "I use LM Studio with MLX support, which is specifically faster on M series Macs.", "excerpt": "Yes, I use LM Studio with MLX support, which is specifically faster on M series Macs. I am not sure if llama is the same, but I guess what I\u2019m saying is if you want the performance to be good on M series you have to use models packaged in the right format.", "context": "M series Mac user", "use_case": "running local models on Mac", "switched_from": "", "story": "Bonsai 2 27B: Near-Lossless Compression in a 9x Smaller Footprint"}, {"id": 796018064, "item_id": 49751129, "url": "https://news.ycombinator.com/item?id=49751129", "by": "Pragmata", "date": "2026-09-18", "stance": "neutral", "quote": "offload some into system ram when using LM studio", "excerpt": "I got a RTX5080 as primary and a RTX3050 (8gb) as secondary. I always max out on context size so that means i need to split the model between both gpus + offload some into system ram when using LM studio. I'm guessing that is what accounts for most of the slowdown when i use that. With this recipe i don't have to worry about it. It's pretty fast, and since its all running in the 5080 with max context it's stable (tho", "context": "", "use_case": "running local LLMs", "switched_from": "", "story": "Run QWEN3.8 27B on 16gb Nvidia GPUs"}, {"id": 795911664, "item_id": 49744479, "url": "https://news.ycombinator.com/item?id=49744479", "by": "selicos", "date": "2026-09-17", "stance": "recommend", "quote": "Using: LM Studio (primary), Lemonade", "excerpt": "So I have to connect over MCP or log into github through the page to share anything? How about being able to post a markdown summary of my setup instead of plugging my credentials and tools into a third party site? Strix HALO 128gb - Framework mainboard in a custom SFF PC. Just moved from Ubuntu to Fedora 44. Using: LM Studio (primary), Lemonade, not yet got into vLLM and llama.cpp directyly after moving to Fedora. R", "context": "infra engineer, local LLM tinkerer", "use_case": "running local LLMs", "switched_from": "", "story": "Show HN: Share your AI Setup, Learn from others"}, {"id": 793011683, "item_id": 49563230, "url": "https://news.ycombinator.com/item?id=49563230", "by": "KronisLV", "date": "2026-09-04", "stance": "neutral", "quote": "I've also heard okay things about LM Studio", "excerpt": "> Download ollama and give it a try Weren't people pissed off at them for a variety of reasons? Not giving attribution etc. In case anyone wants other good alternatives, there's llama.cpp and vLLM, each harder to setup but opens the door for squeezing more performance out of your hardware. I've also heard okay things about LM Studio and I think Unsloth had their own thing as well: https://unsloth.ai/docs/new/studio >", "context": "", "use_case": "running local LLMs", "switched_from": "", "story": "Which tools do Claude, Codex and Cursor choose? We measured 17k runs to find out"}, {"id": 792106146, "item_id": 49506634, "url": "https://news.ycombinator.com/item?id=49506634", "by": "Schlagbohrer", "date": "2026-08-31", "stance": "recommend", "quote": "with Qwen3.8-27B in back in LM Studio, to be amazing and i use it daily", "excerpt": "I have found the Pi agent harness, running in a docker container, with Qwen3.8-27B in back in LM Studio, to be amazing and i use it daily. Just last weekend I used it to do PDF editing which previously would have required me to download really shady software online to do. I have never tried OpenClaw so maybe I am missing out but considering how much my agent can modify the Pi environment to work better, I am happy wi", "context": "", "use_case": "local model serving", "switched_from": "", "story": "OpenClaw 2.0, Accidentally"}, {"id": 791772320, "item_id": 49485770, "url": "https://news.ycombinator.com/item?id=49485770", "by": "wolvoleo", "date": "2026-08-29", "stance": "recommend", "quote": "I've heard good thing about LM Studio", "excerpt": "I'm not sure. I don't use Mac anymore. It used to be my daily driver but they pushed me away a few years ago with the constant iOSification. I've heard good thing about LM Studio, that's about it. https://lmstudio.ai/ I just run a server with Linux (previously multiple servers but I found a way to add multiple GPUs to a single one).", "context": "former Mac user, now Linux", "use_case": "running local LLMs on macOS", "switched_from": "", "story": "Run Qwen3.8 27B locally: real numbers from my Mac Studio"}, {"id": 791703075, "item_id": 49481442, "url": "https://news.ycombinator.com/item?id=49481442", "by": "wolvoleo", "date": "2026-08-28", "stance": "compare", "quote": "I guess you can get it a lot more click to go with something like LM Studio though but I want to use it on a server", "excerpt": "Similar here, except I used Opus 4.8. I don't mind using cloud models to make code as long as I don't put actual private data into the model. And I can't run qwen 27b reliably at the moment, I need a better gpu. It was pretty easy mode like this tbh, though you do need to know what you're doing. I use it with llama-server and openwebui. I guess you can get it a lot more click to go with something like LM Studio thoug", "context": "", "use_case": "", "switched_from": "", "story": "Run Qwen3.8 27B locally: real numbers from my Mac Studio"}, {"id": 790955507, "item_id": 49434719, "url": "https://news.ycombinator.com/item?id=49434719", "by": "jlokier", "date": "2026-08-25", "stance": "neutral", "quote": "Others use LM Studio, oMLX, etc.", "excerpt": "You can't run Linux directly on these. Asahi Linux supports up to M2 only. Linux runs very well in a VM on macOS. There are many good options for this, some free and open source (QEMU, UTM, Lima, Colima), some proprietary (VMware Fusion, Parallels). But Linux in a VM doesn't get access to the real GPU, so model performance is limited. Those running on the CPU perform well, and those needing the GPU don't. However, ma", "context": "", "use_case": "local model runner", "switched_from": "", "story": "Apple introduces M6 and M5 Ultra"}, {"id": 788820531, "item_id": 49301283, "url": "https://news.ycombinator.com/item?id=49301283", "by": "jwr", "date": "2026-08-14", "stance": "warn", "quote": "I got tired of LM Studio not implementing the image-related parameters", "excerpt": "I have an M4 Max (unfortunately 64GB). I have been running the Qwen 35B A3B one for a while now, after testing and benchmarking a number of models. That one was consistently the best in class for tasks like despamming, E-mail classification, OCR and dictation post-processing. It was also really fast (90 tokens/s). I'm benchmarking the 3.8 model now, it seems it is better (near-perfect score on my E-mail spam filterin", "context": "", "use_case": "running local models", "switched_from": "", "story": "Qwen3.8-27B"}, {"id": 788296289, "item_id": 49268518, "url": "https://news.ycombinator.com/item?id=49268518", "by": "imrehg", "date": "2026-08-11", "stance": "warn", "quote": "Their ROCm runtime does not enable the the AMD GPU inside the Framework", "excerpt": "llama.cpp works pretty well for me on the Framework 13 laptop, but the current era of \"move fast, break things, rarely fix\" (sorry, that's how it feels), bites here quite a bit. Two examples: - https://github.com/ggml-org/llama.cpp/pull/25863 Someone's few lines change broke the native (ROCm) support for the AMD GPU inside Framework (and other integrated systems), and any rollback or proper fix is pending for almost ", "context": "", "use_case": "running GGUF models", "switched_from": "", "story": "llama.cpp"}, {"id": 788180289, "item_id": 49261268, "url": "https://news.ycombinator.com/item?id=49261268", "by": "simonw", "date": "2026-08-11", "stance": "neutral", "quote": "I used LM Studio and this model file", "excerpt": "Runs fast on my Mac - ~100 tokens/second - but it's a bit of an over-thinker. It sketched out four different SVGs in the reasoning trace before returning a (bad) pelican riding a bicycle: https://tools.simonwillison.net/markdown-svg-renderer#url=https%3A%2F%2Fgist.github.com%2Fsimonw%2Fabadd7293438a97eef76d03916659335 I used LM Studio and this model file: https://lmstudio.ai/models/nvidia/nemotron-3.5-lightning I got", "context": "", "use_case": "run local models", "switched_from": "", "story": "Nvidia Nemotron 3.5 Lightning"}, {"id": 787906129, "item_id": 49244133, "url": "https://news.ycombinator.com/item?id=49244133", "by": "DanielHB", "date": "2026-08-10", "stance": "recommend", "quote": "I eventually switched to LM studio and the same model runs much better, like 70tk/s.", "excerpt": "I have tried running llamma.cpp on my PC and I found it hard getting it to run at decent speed. On Qwen 3.5-9b I get at most 10tk/s. I eventually switched to LM studio and the same model runs much better, like 70tk/s. Not sure if it was because I was running llama.cpp inside podman or badly tuned LLM arguments. But LM studio is unfortunately much more practical. Although I agree with you. I do not really know what ki", "context": "", "use_case": "running local LLMs", "switched_from": "llama.cpp", "story": "DeepSeek costs OpenCode Go user $1.14/day; dual DGX breaks even in 24 years"}, {"id": 787762067, "item_id": 49235129, "url": "https://news.ycombinator.com/item?id=49235129", "by": "thread123", "date": "2026-08-09", "stance": "neutral", "quote": "I had a lot of chats in OpenRouter, google AI Studio and lm studio", "excerpt": "I had a lot of chats in OpenRouter, google AI Studio and lm studio and no good way to view/search/combine them together. ThreadShelf started as a local archive with semantic search/MCP, and later grew support for continuing conversations through llama.cpp or OpenRouter. https://github.com/ChrystianSchutz/ThreadShelf It is my first time creating oss so please be gentle.", "context": "", "use_case": "local LLM chats", "switched_from": "", "story": "Ask HN: What are you working on? (August 2026)"}, {"id": 787399440, "item_id": 49212465, "url": "https://news.ycombinator.com/item?id=49212465", "by": "joe-dakroub", "date": "2026-08-07", "stance": "recommend", "quote": "Bionic provides a nice experience and makes trying new models out trivial.", "excerpt": "I use Bionic (https://lmstudio.ai) and various local models like Qwen Coder Next to work on various tasks. Honestly, none of the models that are runnable on a typical MacBook Pro come close to the cloud-based frontier models (open or closed) but Bionic provides a nice experience and makes trying new models out trivial.", "context": "", "use_case": "running local models", "switched_from": "", "story": "Ask HN: How is everyone using Local LLMs?"}, {"id": 787288384, "item_id": 49205524, "url": "https://news.ycombinator.com/item?id=49205524", "by": "alasr", "date": "2026-08-06", "stance": "recommend", "quote": "\"LM Studio\"[2] is not bad to start with as you'll be able to download, start/stop/manage and chat with your LLM model all in one place", "excerpt": "- See \"Llama.cpp vs Ollama\"[0] and \"Llama.cpp vs LM Studio\"[1] for a high-level features comparison and find out which one suite you best (check out the \"Target Users\" category[0][1], at least). - If you're still not sure, \"LM Studio\"[2] is not bad to start with as you'll be able to download, start/stop/manage and chat with your LLM model all in one place: a single desktop app. Also, once installed, enable \"Developer", "context": "", "use_case": "running local LLMs", "switched_from": "", "story": "Qwen3.8-Max: A New Bar for Coding and Cowork"}, {"id": 786509616, "item_id": 49156851, "url": "https://news.ycombinator.com/item?id=49156851", "by": "imrehg", "date": "2026-08-03", "stance": "warn", "quote": "Case to point is LM Studio. It's closed source, has bugs (duh!), and there's at least a GitHub issue tracker to report the bugs -- but then by and large nothing happens", "excerpt": "I have sent countless small bug fixes even before, to tools that I use that I could dig into when something was off... No big heavy lifting, but plenty of \"tinker on your tools\" stuff. And I don't think what I did was that special, so maybe the experiences are different? The current LLM-driven stuff seems to break down the expectations, and now there are a lot of places which just ignore anything I send in (the same ", "context": "contributes small bug fixes to tools he uses", "use_case": "running local LLMs", "switched_from": "", "story": "Devtools must be open source"}, {"id": 785593648, "item_id": 49099603, "url": "https://news.ycombinator.com/item?id=49099603", "by": "ktosobcy", "date": "2026-07-29", "stance": "recommend", "quote": "Recently I started playing with LM Studio and local models", "excerpt": "Recently I started playing with LM Studio and local models and found out that `gemma-4-26b-a4b` is suprisingly capable. I don't need elaborate akin to \"create complete app to do X\" or \"refactor the whole codebase of bazzilion of LOC\" but rather \"how to go about doing thing X\" or for language study (explaining nuances of phrasal verbs or subtelties of vocabulary in other languagues) and darn -- the results are rather ", "context": "", "use_case": "running local models", "switched_from": "", "story": "Self-hosting Kimi K3: 20% more hardware cost, 20% better task resolution"}, {"id": 783823969, "item_id": 48988998, "url": "https://news.ycombinator.com/item?id=48988998", "by": "jsomedon", "date": "2026-07-21", "stance": "compare", "quote": "I can't really tell why would I use this over like lm studio, jan.ai and such?", "excerpt": "I can't really tell why would I use this over like lm studio, jan.ai and such?", "context": "", "use_case": "running local models", "switched_from": "", "story": "Nativ: Run frontier open models locally on your Mac"}, {"id": 783782145, "item_id": 48986384, "url": "https://news.ycombinator.com/item?id=48986384", "by": "dofm", "date": "2026-07-20", "stance": "recommend", "quote": "LM Studio (nicer app on the Mac)", "excerpt": "On an M1 Max I have been using either Unsloth Studio (which is basically a web app) or LM Studio (nicer app on the Mac). You can use the Google AI Edge Gallery to play with the smaller Gemma models (but at the moment the QAT variants don't seem to be there unless I am missing something). I think it's likely the 26B QAT model won't fit in your machine \u2014 you may be able to fit one of the UD_Q3 or UD_Q2 variants but whe", "context": "M1 Max user", "use_case": "running local models", "switched_from": "", "story": "Nativ: Run frontier open models locally on your Mac"}, {"id": 783755361, "item_id": 48984710, "url": "https://news.ycombinator.com/item?id=48984710", "by": "erikkahler", "date": "2026-07-20", "stance": "recommend", "quote": "the excellent LM Studio", "excerpt": "To clarify, this MIT-licensed app is from the very same dev, 'Prince Canuma', who maintains the popular MLX-VLM library (https://github.com/Blaizzy/mlx-vlm). MLX-VLM is a long-time dependency of the excellent LM Studio and others because it can provide faster inference on Apple devices than llama.cpp. Historically, MLX is a smaller community than CUDA, but has some of the fastest updates upon the release of new model", "context": "", "use_case": "local LLM runner", "switched_from": "", "story": "Nativ: Run frontier open models locally on your Mac"}, {"id": 783585728, "item_id": 48974108, "url": "https://news.ycombinator.com/item?id=48974108", "by": "Grombobulous", "date": "2026-07-19", "stance": "neutral", "quote": "I was already using LMStudio", "excerpt": "This article is really helpful for finding alternatives. I was already using LMStudio but I love that I now know about a couple of true open source alternatives I can try out.", "context": "", "use_case": "running local models", "switched_from": "", "story": "Ollama: All Aboard Open Models"}, {"id": 783137232, "item_id": 48946077, "url": "https://news.ycombinator.com/item?id=48946077", "by": "hypercube33", "date": "2026-07-17", "stance": "recommend", "quote": "I love LM studio and this is super cool", "excerpt": "I love LM studio and this is super cool and I plan on trying it out tonight. I'm absolutely dying for an android link app and I'm jealous of iPhone users", "context": "", "use_case": "running local models", "switched_from": "", "story": "LM Studio Bionic: the AI agent for open models"}, {"id": 783094080, "item_id": 48943380, "url": "https://news.ycombinator.com/item?id=48943380", "by": "IronWolve", "date": "2026-07-16", "stance": "recommend", "quote": "I love lm-studio, so cant wait to see how this goes. For local I normally use opencode + lmstudio.", "excerpt": "It was fun to try out which local models made better agents. Few thoughts. * locked to a single dir, so no system wide access. * no local web search, can be fixed ddg or local mcp. * no ssh, I want to have it ssh into my server and do the work. * doesnt show the model being loaded, needs a bar/% counter. * Can you drag/drop documents in the work dirs, or only + add them? I love lm-studio, so cant wait to see how this", "context": "", "use_case": "running local models", "switched_from": "", "story": "LM Studio Bionic: the AI agent for open models"}, {"id": 783079281, "item_id": 48942455, "url": "https://news.ycombinator.com/item?id=48942455", "by": "inventor7777", "date": "2026-07-16", "stance": "recommend", "quote": "I really love LM Studio", "excerpt": "I have never previously tried a agentic harness for local models yet, but I really love LM Studio so I gave Bionic a shot immediately after reading this! First impression: it works great. I use Codex as my main agent, and the UI looks similar enough that it's familiar and simple to get started. I just pointed it to my existing LM Studio models library, ran Qwen3.6 35B, and the results are exactly what I would hope fo", "context": "", "use_case": "running local models", "switched_from": "", "story": "LM Studio Bionic: the AI agent for open models"}, {"id": 783064624, "item_id": 48941539, "url": "https://news.ycombinator.com/item?id=48941539", "by": "codazoda", "date": "2026-07-16", "stance": "recommend", "quote": "which is part of the reason I just switched to LM Studio from Ollama.", "excerpt": "I\u2019m worried about the switch in business model here, which is part of the reason I just switched to LM Studio from Ollama. > use the largest frontier open source models through LM Studio Secure Cloud", "context": "", "use_case": "running local models", "switched_from": "Ollama", "story": "LM Studio Bionic: the AI agent for open models"}, {"id": 782808688, "item_id": 48925543, "url": "https://news.ycombinator.com/item?id=48925543", "by": "rvba", "date": "2026-07-15", "stance": "neutral", "quote": "I use LM studio (local server), but can switch to a different tool.", "excerpt": "Thank you for your reply. I use LM studio (local server), but can switch to a different tool. Do you know how to switch it in LM studio?", "context": "", "use_case": "serving local models", "switched_from": "", "story": "Running Gemma 4 26B at 5 tokens/SEC on a 13-year-old Xeon with no GPU"}, {"id": 781698784, "item_id": 48856174, "url": "https://news.ycombinator.com/item?id=48856174", "by": "kamranjon", "date": "2026-07-09", "stance": "recommend", "quote": "I use LMStudio pretty regularly", "excerpt": "I dunno I use LMStudio pretty regularly and the MLX folks and the community usually have MLX versions of new model releases up within a day or two.", "context": "", "use_case": "running local models on Mac", "switched_from": "", "story": "Apple Silicon Exec Explains Mac Mini AI Demand and On-Device Future"}, {"id": 780235248, "item_id": 48764703, "url": "https://news.ycombinator.com/item?id=48764703", "by": "kamranjon", "date": "2026-07-02", "stance": "recommend", "quote": "I usually use LM Studio for this type of thing", "excerpt": "I'm really happy this is one of the top comments here, I am fully local as well. Just wanted to leave a note for folks who might not have the memory to run a big 32gb model - I just found out there are some pruned models that have really good performance and If I had a smaller machine I might try this pruned unsloth Q4 quant of GLM 4.7 flash that sits at 14gb: https://huggingface.co/unsloth/GLM-4.7-Flash-REAP-23B-A3B", "context": "runs fully local", "use_case": "running local models", "switched_from": "", "story": "Kimi K2.7 Code is generally available in GitHub Copilot"}, {"id": 779825617, "item_id": 48739101, "url": "https://news.ycombinator.com/item?id=48739101", "by": "razster", "date": "2026-06-30", "stance": "neutral", "quote": "I may have to switch away from LM-studio if I can improve token speed.", "excerpt": "Thanks for sharing that. I have the same card but 96gb ram. I use PI.dev to connect to LM-Studio. I may have to switch away from LM-studio if I can improve token speed. I think I range from 32-40t/s. qwen3.6-35b-a3b-genesis-v2-apex-mtp.", "context": "has 12GB 4070 with 96GB RAM", "use_case": "running local models", "switched_from": "", "story": "Qwen 3.6 27B is the sweet spot for local development"}, {"id": 779719552, "item_id": 48732472, "url": "https://news.ycombinator.com/item?id=48732472", "by": "Saris", "date": "2026-06-30", "stance": "neutral", "quote": "The main thing in LM studio (or whatever software you use", "excerpt": "I run it on a 12GB 4070 with 32GB system RAM. 35B A4B means only part of the model is active at a time so it takes a lot less VRAM than a dense 35B model would. The main thing in LM studio (or whatever software you use, assuming it has fairly up to date stuff and exposes the toggles) is to offload MoE layers to the CPU, and use K/V cache quantization at Q8_0 or Q4_0. Since you have more VRAM than I do, you could prob", "context": "12GB 4070, 32GB RAM", "use_case": "running local MoE models", "switched_from": "", "story": "Qwen 3.6 27B is the sweet spot for local development"}, {"id": 779568064, "item_id": 48723004, "url": "https://news.ycombinator.com/item?id=48723004", "by": "dannyw", "date": "2026-06-29", "stance": "neutral", "quote": "I can just hit my own APIs (via LM Studio) on the MBP over Tailscale.", "excerpt": "I purchased mine for approximately $4400 AUD before the price hikes. That unit is now ~$5100 AUD. I use my MBP essentially as my workstation, it's almost always plugged in. I have a MBA (M4, 24GB RAM) that I picked up for ~A$1500 or so, and that's an amazing daily driver. I don't do local LLM inference on that unit, I can just hit my own APIs (via LM Studio) on the MBP over Tailscale.", "context": "", "use_case": "local LLM API", "switched_from": "", "story": "Qwen 3.6 27B is the sweet spot for local development"}, {"id": 778893924, "item_id": 48680870, "url": "https://news.ycombinator.com/item?id=48680870", "by": "kylenessen", "date": "2026-06-25", "stance": "recommend", "quote": "I use LMStudio which supports OpenAI and Anthropic compatible api endpoints", "excerpt": "I really wanted to like this, but unfortunately couldn't see how it improves my experience over Obsidian or VS Code. The fact that I have to juggle between OpenKnowledge and Codex to engage the AI, while also accepting a barebones Obsidian, is a real bummer. From what I can tell, you are saving me a few key strokes with moving prompts around. What I really want is the AI to live IN the app, like VS Code, and then mov", "context": "", "use_case": "running local models", "switched_from": "", "story": "Show HN: OpenKnowledge \u2013 open-source alternative to Obsidian/Notion"}, {"id": 778229808, "item_id": 48639363, "url": "https://news.ycombinator.com/item?id=48639363", "by": "mapontosevenths", "date": "2026-06-22", "stance": "neutral", "quote": "I think he just uses LMStudio most of the time on his", "excerpt": "I don't know. I think he just uses LMStudio most of the time on his, but that's one place I can say the spark really shines for me. I'm a Linux guy, but also don't always have alot of time. The Spark comes out of the box with a nice Linux distro that's pre-configured to be easy to setup and the guides and online resources make getting up and running trivial, for even some complex tasks. You would have to do a LOT of ", "context": "", "use_case": "running local models", "switched_from": "", "story": "Unsloth GLM-5.2 \u2013 How to Run Locally"}, {"id": 777788913, "item_id": 48611807, "url": "https://news.ycombinator.com/item?id=48611807", "by": "anonym29", "date": "2026-06-20", "stance": "recommend", "quote": "llama.cpp supports structured outputs, as do some more consumer-oriented tools that wrap like LM Studio.", "excerpt": "In that case, you don't want to be using Claude Code, which is more of a consumer product; you instead want to control the inference stack yourself. What you are looking for is structured output (you give the inference engine a JSON schema you define that the response must conform to) + a JSON schema validator that parses the output, checks if the response is valid JSON. If it is, you're good to go, if not, run the i", "context": "", "use_case": "structured output inference", "switched_from": "", "story": "Ask HN: What technique do you use to make Claude Code deterministic?"}, {"id": 776911824, "item_id": 48556989, "url": "https://news.ycombinator.com/item?id=48556989", "by": "simonw", "date": "2026-06-16", "stance": "recommend", "quote": "I strongly recommend trying LM Studio - it's the lowest friction way to try out models", "excerpt": "I strongly recommend trying LM Studio - it's the lowest friction way to try out models, you can browse https://lmstudio.ai/models and click \"Get\" and then \"Run in LM Studio\" to download and run a model. With 96GB I'd start with the Gemma 4 and Qwen 3.6 models. Any of those should work fine.", "context": "", "use_case": "running local models", "switched_from": "", "story": "Running local models is good now"}, {"id": 776766930, "item_id": 48547933, "url": "https://news.ycombinator.com/item?id=48547933", "by": "mgsram", "date": "2026-06-15", "stance": "neutral", "quote": "with open code as the harness and llmster(LM Studio)", "excerpt": "I have been using local LLMs for about a year and I have settled now on Qwen3.6 27b dense model in GGUF on Mac Studio with 512G of RAM with open code as the harness and llmster(LM Studio). I have also used the Qwen 3.6 35B-A3B but the dense model's accuracy is next level with the tradeoff being tokens/sec. With the Qwen3.6 27b, I usually get anywhere from 25-40 tokens/second. Initially I used them for simple tools bu", "context": "", "use_case": "running local models", "switched_from": "", "story": "Ask HN: Has anyone replaced Claude/GPT with a local model for daily coding?"}, {"id": 776756881, "item_id": 48547305, "url": "https://news.ycombinator.com/item?id=48547305", "by": "heisenbit", "date": "2026-06-15", "stance": "neutral", "quote": "I use LM Studio to run and am using it via VSCode.", "excerpt": "I think it is work to set up but I'm also learning a lot setting it up. Mainly using qwen/qwen3.6-35b-a3b mlx with my 48GB M4 MBP which leaves me just enough headroom for docker dev-container and other basics. I use LM Studio to run and am using it via VSCode. A big difference made the system prompt improving the tool integration (I asked GPT for guidance on that). Before that it was not making changes but regenerati", "context": "", "use_case": "running local models", "switched_from": "", "story": "Ask HN: Has anyone replaced Claude/GPT with a local model for daily coding?"}, {"id": 776215408, "item_id": 48513463, "url": "https://news.ycombinator.com/item?id=48513463", "by": "hmontazeri", "date": "2026-06-12", "stance": "recommend", "quote": "I use LM Studio with the local server it ships and connect it to opencode. Takes 2 min to setup", "excerpt": "I use LM Studio with the local server it ships and connect it to opencode. Takes 2 min to setup", "context": "", "use_case": "local model server", "switched_from": "", "story": "How to Setup a Local Coding Agent on macOS"}, {"id": 771593889, "item_id": 48224618, "url": "https://news.ycombinator.com/item?id=48224618", "by": "satvikpendem", "date": "2026-05-21", "stance": "compare", "quote": "open source alternative to the more widely known LM Studio", "excerpt": "Unsloth Studio [0] is what I recommend these days, open source alternative to the more widely known LM Studio, and also built by the people who make good quantizations of released models. With MTP support not merged in you should get 2x token generation speed with no accuracy difference. They also have MLX quants if you scroll down a bit, which is a format specifically for macOS' Metal GPU acceleration but that's not", "context": "", "use_case": "running local models", "switched_from": "", "story": "Indexing a year of video locally on a 2021 MacBook with Gemma4-31B (50GB swap)"}, {"id": 757880080, "item_id": 47367505, "url": "https://news.ycombinator.com/item?id=47367505", "by": "brcmthrowaway", "date": "2026-03-13", "stance": "recommend", "quote": "LM Studio has a server that serves models over HTTP (localhost).", "excerpt": "Basically LM Studio has a server that serves models over HTTP (localhost). Configure/enable the server and connect OpenCode to it. Try this article https://advanced-stack.com/fields-notes/qwen35-opencode-lm-studio-agentic-coding-on-m1.html I'm looking for an alternative to OpenCode though, I can barely see the UI.", "context": "", "use_case": "local model server", "switched_from": "", "story": "Can I Run AI locally?"}, {"id": 747300482, "item_id": 46706280, "url": "https://news.ycombinator.com/item?id=46706280", "by": "river_otter", "date": "2026-01-21", "stance": "neutral", "quote": "I'm documenting my experience using aoe + OpenCode + LM Studio + GLM-4.7 Flash + Mac Mini M4Pro 64GB Mem", "excerpt": "Excited to finally be able to give this a try today. I'm documenting my experience using aoe + OpenCode + LM Studio + GLM-4.7 Flash + Mac Mini M4Pro 64GB Mem on this thread if anyone wants to follow along and or give me advice about how badly I'm messing up the settings https://x.com/natebrake/status/2013978241573204246 Thus far, the 6-bit quant MLX weights were too much and crashed LMS with OOM", "context": "", "use_case": "running local models", "switched_from": "", "story": "GLM-4.7-Flash"}]}