{"slug": "muse-glimmer", "name": "Muse Glimmer", "category": "ai", "category_name": "AI models and assistants", "description": "Meta's open agentic 30B-parameter language model that runs locally.", "recommenders": 5, "commenters": 6, "warners": 0, "score": 4.959, "rising": 5.0, "first": "2026-08-10", "latest": "2026-09-14", "why": "HN commenters like it for producing decent results while running locally at 30B params, and for using tokens efficiently enough to finish tasks quicker than A3B.", "why_is_quote": false, "has_verdict": true, "sentiment": {"recommend": 5, "neutral": 0, "compare": 1, "warn": 0}, "spark": [0, 0, 0, 0, 0, 0, 0, 0, 0, 5, 0, 0], "listed": true, "loved": ["Output quality", "Token efficiency"], "caveat": "Slow generation", "short_quote": "I am getting good results with muse-glimmer running locally, with the caveat that everything runs slowly", "icon": true, "domain": "meta.ai", "github_repo": null, "official_url": "https://meta.ai", "pricing": "unknown", "aliases": ["Muse Glimmer", "muse-glimmer"], "verdict": [{"text": "HN commenters like it for producing decent results while running locally at 30B params", "cites": [{"id": 788199792, "item_id": 49262487, "url": "https://news.ycombinator.com/item?id=49262487", "by": "joe-dakroub", "date": "2026-08-11", "stance": "recommend", "quote": "runs locally at 30B params. Slow for most things but it does produce decent results", "excerpt": "Meta just released Muse Glimmer that runs locally at 30B params. Slow for most things but it does produce decent results: https://research.meta.ai/blog/introducing-muse-glimmer-open-agentic-model", "context": "", "use_case": "local coding model", "switched_from": "", "story": "Ask HN: Claude Code Alternatives"}, {"id": 787915186, "item_id": 49244699, "url": "https://news.ycombinator.com/item?id=49244699", "by": "mark_l_watson", "date": "2026-08-10", "stance": "recommend", "quote": "I am getting good results with muse-glimmer running locally, with the caveat that everything runs slowly", "excerpt": "Meta is rocking AI. As of last week I have been using their excellent muse coding harness with their model Muse Spark 1.2. Starting this morning I am running their new local 30B model muse-glimmer on my old MacMini 32G using Ollama (remember to increase the context size!) and pi coding harness. I am getting good results with muse-glimmer running locally, with the caveat that everything runs slowly (e.g., give it a ta", "context": "runs on an old MacMini 32G", "use_case": "local coding model on Mac mini", "switched_from": "", "story": "Meta Muse Glimmer \u2013 open weights 30B local coding model"}, {"id": 787977648, "item_id": 49248603, "url": "https://news.ycombinator.com/item?id=49248603", "by": "dofm", "date": "2026-08-10", "stance": "recommend", "quote": "I am enjoying testing Muse Glimmer \u2014 it's really quite impressive on chat", "excerpt": "One thing I would caution is staying out of the prediction market like this. Tech tends to get boring when you judge current products against the hypothetical capabilities of unannounced products that may never ship. It's like comparing Nikon cameras against Canon camera rumours, or comparing iPhones against unannounced and therefore largely imaginary Samsungs. - If they do a Qwen 3.8 35B A3B (and I hope they do beca", "context": "", "use_case": "local chat and agentic use", "switched_from": "", "story": "Meta Muse Glimmer \u2013 open weights 30B local coding model"}]}, {"text": ", and for using tokens efficiently enough to finish tasks quicker than A3B.", "cites": [{"id": 791818481, "item_id": 49488655, "url": "https://news.ycombinator.com/item?id=49488655", "by": "solarkraft", "date": "2026-08-29", "stance": "recommend", "quote": "it  uses them very efficiently, caveman style", "excerpt": "NGL: I don\u2019t really have a good way to find out right now. It also doesn\u2019t matter that much because the way the models use the tokes varies a lot. Qwen 3.8 is known for overthinking while Muse Glimmer may be a little slower per token, but it uses them very efficiently, caveman style. Generation speed isn\u2019t the bottleneck anyway, at least on pre M4/M5 devices (the newer chips got significant processing acceleration). ", "context": "", "use_case": "local LLM", "switched_from": "", "story": "GLM-5.3 is now open-weight"}, {"id": 789218912, "item_id": 49326182, "url": "https://news.ycombinator.com/item?id=49326182", "by": "bitexploder", "date": "2026-08-16", "stance": "recommend", "quote": "I have been using Muse Glimmer for a few days instead of A3B. It gets the job done quicker than A3B despite being several times slower.", "excerpt": "I have been using Muse Glimmer for a few days instead of A3B. It gets the job done quicker than A3B despite being several times slower.", "context": "", "use_case": "local model tasks", "switched_from": "A3B", "story": "Qwen 3.8 27B is excellent, but it defaults to overthinking things"}]}, {"text": " The most common complaint is that it runs slowly", "cites": [{"id": 791818481, "item_id": 49488655, "url": "https://news.ycombinator.com/item?id=49488655", "by": "solarkraft", "date": "2026-08-29", "stance": "recommend", "quote": "it  uses them very efficiently, caveman style", "excerpt": "NGL: I don\u2019t really have a good way to find out right now. It also doesn\u2019t matter that much because the way the models use the tokes varies a lot. Qwen 3.8 is known for overthinking while Muse Glimmer may be a little slower per token, but it uses them very efficiently, caveman style. Generation speed isn\u2019t the bottleneck anyway, at least on pre M4/M5 devices (the newer chips got significant processing acceleration). ", "context": "", "use_case": "local LLM", "switched_from": "", "story": "GLM-5.3 is now open-weight"}, {"id": 789218912, "item_id": 49326182, "url": "https://news.ycombinator.com/item?id=49326182", "by": "bitexploder", "date": "2026-08-16", "stance": "recommend", "quote": "I have been using Muse Glimmer for a few days instead of A3B. It gets the job done quicker than A3B despite being several times slower.", "excerpt": "I have been using Muse Glimmer for a few days instead of A3B. It gets the job done quicker than A3B despite being several times slower.", "context": "", "use_case": "local model tasks", "switched_from": "A3B", "story": "Qwen 3.8 27B is excellent, but it defaults to overthinking things"}, {"id": 788199792, "item_id": 49262487, "url": "https://news.ycombinator.com/item?id=49262487", "by": "joe-dakroub", "date": "2026-08-11", "stance": "recommend", "quote": "runs locally at 30B params. Slow for most things but it does produce decent results", "excerpt": "Meta just released Muse Glimmer that runs locally at 30B params. Slow for most things but it does produce decent results: https://research.meta.ai/blog/introducing-muse-glimmer-open-agentic-model", "context": "", "use_case": "local coding model", "switched_from": "", "story": "Ask HN: Claude Code Alternatives"}, {"id": 787915186, "item_id": 49244699, "url": "https://news.ycombinator.com/item?id=49244699", "by": "mark_l_watson", "date": "2026-08-10", "stance": "recommend", "quote": "I am getting good results with muse-glimmer running locally, with the caveat that everything runs slowly", "excerpt": "Meta is rocking AI. As of last week I have been using their excellent muse coding harness with their model Muse Spark 1.2. Starting this morning I am running their new local 30B model muse-glimmer on my old MacMini 32G using Ollama (remember to increase the context size!) and pi coding harness. I am getting good results with muse-glimmer running locally, with the caveat that everything runs slowly (e.g., give it a ta", "context": "runs on an old MacMini 32G", "use_case": "local coding model on Mac mini", "switched_from": "", "story": "Meta Muse Glimmer \u2013 open weights 30B local coding model"}]}, {"text": ", and one commenter found it changes its position too easily compared with Qwen.", "cites": [{"id": 795045377, "item_id": 49690336, "url": "https://news.ycombinator.com/item?id=49690336", "by": "noir_lord", "date": "2026-09-14", "stance": "compare", "quote": "Muse came close but what impressed me about Qwen is it\u2019ll push back", "excerpt": "I did, Muse came close but what impressed me about Qwen is it\u2019ll push back if it thinks it\u2019s right even when it isn\u2019t, I can work with that, Muse tended to flip between states too easily/too much. It is a good model but Qwen (at least for the things I use it for) edges just ahead, it seems much better at the \u201crip this apart, suggest improvements, touch nothing\u201d use case where I can use it as a second set of eyes, I d", "context": "", "use_case": "local code review", "switched_from": "", "story": "Astra and Fable still hack on simple variants of alignment evals from 2025"}]}], "praise": [{"label": "Output quality", "summary": "Commenters report decent results locally and impressive chat.", "quotes": [{"id": 788199792, "item_id": 49262487, "url": "https://news.ycombinator.com/item?id=49262487", "by": "joe-dakroub", "date": "2026-08-11", "stance": "recommend", "quote": "runs locally at 30B params. Slow for most things but it does produce decent results", "excerpt": "Meta just released Muse Glimmer that runs locally at 30B params. Slow for most things but it does produce decent results: https://research.meta.ai/blog/introducing-muse-glimmer-open-agentic-model", "context": "", "use_case": "local coding model", "switched_from": "", "story": "Ask HN: Claude Code Alternatives"}, {"id": 787915186, "item_id": 49244699, "url": "https://news.ycombinator.com/item?id=49244699", "by": "mark_l_watson", "date": "2026-08-10", "stance": "recommend", "quote": "I am getting good results with muse-glimmer running locally, with the caveat that everything runs slowly", "excerpt": "Meta is rocking AI. As of last week I have been using their excellent muse coding harness with their model Muse Spark 1.2. Starting this morning I am running their new local 30B model muse-glimmer on my old MacMini 32G using Ollama (remember to increase the context size!) and pi coding harness. I am getting good results with muse-glimmer running locally, with the caveat that everything runs slowly (e.g., give it a ta", "context": "runs on an old MacMini 32G", "use_case": "local coding model on Mac mini", "switched_from": "", "story": "Meta Muse Glimmer \u2013 open weights 30B local coding model"}, {"id": 787977648, "item_id": 49248603, "url": "https://news.ycombinator.com/item?id=49248603", "by": "dofm", "date": "2026-08-10", "stance": "recommend", "quote": "I am enjoying testing Muse Glimmer \u2014 it's really quite impressive on chat", "excerpt": "One thing I would caution is staying out of the prediction market like this. Tech tends to get boring when you judge current products against the hypothetical capabilities of unannounced products that may never ship. It's like comparing Nikon cameras against Canon camera rumours, or comparing iPhones against unannounced and therefore largely imaginary Samsungs. - If they do a Qwen 3.8 35B A3B (and I hope they do beca", "context": "", "use_case": "local chat and agentic use", "switched_from": "", "story": "Meta Muse Glimmer \u2013 open weights 30B local coding model"}]}, {"label": "Token efficiency", "summary": "It uses tokens sparingly, so tasks finish faster despite slower generation.", "quotes": [{"id": 791818481, "item_id": 49488655, "url": "https://news.ycombinator.com/item?id=49488655", "by": "solarkraft", "date": "2026-08-29", "stance": "recommend", "quote": "it  uses them very efficiently, caveman style", "excerpt": "NGL: I don\u2019t really have a good way to find out right now. It also doesn\u2019t matter that much because the way the models use the tokes varies a lot. Qwen 3.8 is known for overthinking while Muse Glimmer may be a little slower per token, but it uses them very efficiently, caveman style. Generation speed isn\u2019t the bottleneck anyway, at least on pre M4/M5 devices (the newer chips got significant processing acceleration). ", "context": "", "use_case": "local LLM", "switched_from": "", "story": "GLM-5.3 is now open-weight"}, {"id": 789218912, "item_id": 49326182, "url": "https://news.ycombinator.com/item?id=49326182", "by": "bitexploder", "date": "2026-08-16", "stance": "recommend", "quote": "I have been using Muse Glimmer for a few days instead of A3B. It gets the job done quicker than A3B despite being several times slower.", "excerpt": "I have been using Muse Glimmer for a few days instead of A3B. It gets the job done quicker than A3B despite being several times slower.", "context": "", "use_case": "local model tasks", "switched_from": "A3B", "story": "Qwen 3.8 27B is excellent, but it defaults to overthinking things"}]}, {"label": "Runs locally", "summary": "The 30B model runs on local hardware, including an older 32GB Mac Mini.", "quotes": [{"id": 788199792, "item_id": 49262487, "url": "https://news.ycombinator.com/item?id=49262487", "by": "joe-dakroub", "date": "2026-08-11", "stance": "recommend", "quote": "runs locally at 30B params. Slow for most things but it does produce decent results", "excerpt": "Meta just released Muse Glimmer that runs locally at 30B params. Slow for most things but it does produce decent results: https://research.meta.ai/blog/introducing-muse-glimmer-open-agentic-model", "context": "", "use_case": "local coding model", "switched_from": "", "story": "Ask HN: Claude Code Alternatives"}, {"id": 787915186, "item_id": 49244699, "url": "https://news.ycombinator.com/item?id=49244699", "by": "mark_l_watson", "date": "2026-08-10", "stance": "recommend", "quote": "I am getting good results with muse-glimmer running locally, with the caveat that everything runs slowly", "excerpt": "Meta is rocking AI. As of last week I have been using their excellent muse coding harness with their model Muse Spark 1.2. Starting this morning I am running their new local 30B model muse-glimmer on my old MacMini 32G using Ollama (remember to increase the context size!) and pi coding harness. I am getting good results with muse-glimmer running locally, with the caveat that everything runs slowly (e.g., give it a ta", "context": "runs on an old MacMini 32G", "use_case": "local coding model on Mac mini", "switched_from": "", "story": "Meta Muse Glimmer \u2013 open weights 30B local coding model"}]}, {"label": "Writing personality", "summary": "One commenter enjoys its fun, slightly snarky writing style.", "quotes": [{"id": 788787328, "item_id": 49299208, "url": "https://news.ycombinator.com/item?id=49299208", "by": "dofm", "date": "2026-08-14", "stance": "recommend", "quote": "the way Muse Glimmer writes. Like a talented, slightly snarky guy who is maybe a bit of a dick but quite fun to be around.", "excerpt": "Claude writes like a guy at a firm I used to work with in the 90s; he was my employer's \"visionary\"; he'd worked at a whole lot of different companies on both sides of the Atlantic in inexplicably high-placed roles given that he was often bluffing, and was considered a lucky hire of a rising star. He'd be called into meetings with high end clients to spout off. He really needed you to know he understood, but very oft", "context": "", "use_case": "writing", "switched_from": "", "story": "Why does Opus 5 feel worse to work with?"}]}], "caveats": [{"label": "Slow generation", "summary": "Several commenters say it is slow per token and slow for most tasks.", "quotes": [{"id": 787915186, "item_id": 49244699, "url": "https://news.ycombinator.com/item?id=49244699", "by": "mark_l_watson", "date": "2026-08-10", "stance": "recommend", "quote": "I am getting good results with muse-glimmer running locally, with the caveat that everything runs slowly", "excerpt": "Meta is rocking AI. As of last week I have been using their excellent muse coding harness with their model Muse Spark 1.2. Starting this morning I am running their new local 30B model muse-glimmer on my old MacMini 32G using Ollama (remember to increase the context size!) and pi coding harness. I am getting good results with muse-glimmer running locally, with the caveat that everything runs slowly (e.g., give it a ta", "context": "runs on an old MacMini 32G", "use_case": "local coding model on Mac mini", "switched_from": "", "story": "Meta Muse Glimmer \u2013 open weights 30B local coding model"}, {"id": 788199792, "item_id": 49262487, "url": "https://news.ycombinator.com/item?id=49262487", "by": "joe-dakroub", "date": "2026-08-11", "stance": "recommend", "quote": "runs locally at 30B params. Slow for most things but it does produce decent results", "excerpt": "Meta just released Muse Glimmer that runs locally at 30B params. Slow for most things but it does produce decent results: https://research.meta.ai/blog/introducing-muse-glimmer-open-agentic-model", "context": "", "use_case": "local coding model", "switched_from": "", "story": "Ask HN: Claude Code Alternatives"}, {"id": 789218912, "item_id": 49326182, "url": "https://news.ycombinator.com/item?id=49326182", "by": "bitexploder", "date": "2026-08-16", "stance": "recommend", "quote": "I have been using Muse Glimmer for a few days instead of A3B. It gets the job done quicker than A3B despite being several times slower.", "excerpt": "I have been using Muse Glimmer for a few days instead of A3B. It gets the job done quicker than A3B despite being several times slower.", "context": "", "use_case": "local model tasks", "switched_from": "A3B", "story": "Qwen 3.8 27B is excellent, but it defaults to overthinking things"}]}, {"label": "Inconsistent stance", "summary": "One commenter found it flips between positions too easily and preferred Qwen for critique.", "quotes": [{"id": 795045377, "item_id": 49690336, "url": "https://news.ycombinator.com/item?id=49690336", "by": "noir_lord", "date": "2026-09-14", "stance": "compare", "quote": "Muse came close but what impressed me about Qwen is it\u2019ll push back", "excerpt": "I did, Muse came close but what impressed me about Qwen is it\u2019ll push back if it thinks it\u2019s right even when it isn\u2019t, I can work with that, Muse tended to flip between states too easily/too much. It is a good model but Qwen (at least for the things I use it for) edges just ahead, it seems much better at the \u201crip this apart, suggest improvements, touch nothing\u201d use case where I can use it as a second set of eyes, I d", "context": "", "use_case": "local code review", "switched_from": "", "story": "Astra and Fable still hack on simple variants of alignment evals from 2025"}]}], "who": [{"by": "mark_l_watson", "context": "runs on an old MacMini 32G", "url": "https://news.ycombinator.com/item?id=49244699", "date": "2026-08-10"}], "tags": ["for local inference", "for coding agents", "for Mac users"], "use_cases": ["local code review", "local llm", "local model tasks", "writing", "local coding model", "local chat and agentic use"], "compared": [{"slug": "claude-code", "name": "Claude Code", "co": 2, "switched_from": 0, "switched_to": 0}], "makers": [], "threads": [{"url": "https://news.ycombinator.com/item?id=49241679", "title": "Meta Muse Glimmer \u2013 open weights 30B local coding model", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=49684393", "title": "Astra and Fable still hack on simple variants of alignment evals from 2025", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49479878", "title": "GLM-5.3 is now open-weight", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49324985", "title": "Qwen 3.8 27B is excellent, but it defaults to overthinking things", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49296740", "title": "Why does Opus 5 feel worse to work with?", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49211265", "title": "Ask HN: Claude Code Alternatives", "mentions": 1}], "sentiment_by_month": {"2025-11": {}, "2025-12": {}, "2026-01": {}, "2026-02": {}, "2026-03": {}, "2026-04": {}, "2026-05": {}, "2026-06": {}, "2026-07": {}, "2026-08": {"recommend": 5}, "2026-09": {"compare": 1}, "2026-10": {}}, "quotes": [{"id": 795045377, "item_id": 49690336, "url": "https://news.ycombinator.com/item?id=49690336", "by": "noir_lord", "date": "2026-09-14", "stance": "compare", "quote": "Muse came close but what impressed me about Qwen is it\u2019ll push back", "excerpt": "I did, Muse came close but what impressed me about Qwen is it\u2019ll push back if it thinks it\u2019s right even when it isn\u2019t, I can work with that, Muse tended to flip between states too easily/too much. It is a good model but Qwen (at least for the things I use it for) edges just ahead, it seems much better at the \u201crip this apart, suggest improvements, touch nothing\u201d use case where I can use it as a second set of eyes, I d", "context": "", "use_case": "local code review", "switched_from": "", "story": "Astra and Fable still hack on simple variants of alignment evals from 2025"}, {"id": 791818481, "item_id": 49488655, "url": "https://news.ycombinator.com/item?id=49488655", "by": "solarkraft", "date": "2026-08-29", "stance": "recommend", "quote": "it  uses them very efficiently, caveman style", "excerpt": "NGL: I don\u2019t really have a good way to find out right now. It also doesn\u2019t matter that much because the way the models use the tokes varies a lot. Qwen 3.8 is known for overthinking while Muse Glimmer may be a little slower per token, but it uses them very efficiently, caveman style. Generation speed isn\u2019t the bottleneck anyway, at least on pre M4/M5 devices (the newer chips got significant processing acceleration). ", "context": "", "use_case": "local LLM", "switched_from": "", "story": "GLM-5.3 is now open-weight"}, {"id": 789218912, "item_id": 49326182, "url": "https://news.ycombinator.com/item?id=49326182", "by": "bitexploder", "date": "2026-08-16", "stance": "recommend", "quote": "I have been using Muse Glimmer for a few days instead of A3B. It gets the job done quicker than A3B despite being several times slower.", "excerpt": "I have been using Muse Glimmer for a few days instead of A3B. It gets the job done quicker than A3B despite being several times slower.", "context": "", "use_case": "local model tasks", "switched_from": "A3B", "story": "Qwen 3.8 27B is excellent, but it defaults to overthinking things"}, {"id": 788787328, "item_id": 49299208, "url": "https://news.ycombinator.com/item?id=49299208", "by": "dofm", "date": "2026-08-14", "stance": "recommend", "quote": "the way Muse Glimmer writes. Like a talented, slightly snarky guy who is maybe a bit of a dick but quite fun to be around.", "excerpt": "Claude writes like a guy at a firm I used to work with in the 90s; he was my employer's \"visionary\"; he'd worked at a whole lot of different companies on both sides of the Atlantic in inexplicably high-placed roles given that he was often bluffing, and was considered a lucky hire of a rising star. He'd be called into meetings with high end clients to spout off. He really needed you to know he understood, but very oft", "context": "", "use_case": "writing", "switched_from": "", "story": "Why does Opus 5 feel worse to work with?"}, {"id": 788199792, "item_id": 49262487, "url": "https://news.ycombinator.com/item?id=49262487", "by": "joe-dakroub", "date": "2026-08-11", "stance": "recommend", "quote": "runs locally at 30B params. Slow for most things but it does produce decent results", "excerpt": "Meta just released Muse Glimmer that runs locally at 30B params. Slow for most things but it does produce decent results: https://research.meta.ai/blog/introducing-muse-glimmer-open-agentic-model", "context": "", "use_case": "local coding model", "switched_from": "", "story": "Ask HN: Claude Code Alternatives"}, {"id": 787977648, "item_id": 49248603, "url": "https://news.ycombinator.com/item?id=49248603", "by": "dofm", "date": "2026-08-10", "stance": "recommend", "quote": "I am enjoying testing Muse Glimmer \u2014 it's really quite impressive on chat", "excerpt": "One thing I would caution is staying out of the prediction market like this. Tech tends to get boring when you judge current products against the hypothetical capabilities of unannounced products that may never ship. It's like comparing Nikon cameras against Canon camera rumours, or comparing iPhones against unannounced and therefore largely imaginary Samsungs. - If they do a Qwen 3.8 35B A3B (and I hope they do beca", "context": "", "use_case": "local chat and agentic use", "switched_from": "", "story": "Meta Muse Glimmer \u2013 open weights 30B local coding model"}, {"id": 787915186, "item_id": 49244699, "url": "https://news.ycombinator.com/item?id=49244699", "by": "mark_l_watson", "date": "2026-08-10", "stance": "recommend", "quote": "I am getting good results with muse-glimmer running locally, with the caveat that everything runs slowly", "excerpt": "Meta is rocking AI. As of last week I have been using their excellent muse coding harness with their model Muse Spark 1.2. Starting this morning I am running their new local 30B model muse-glimmer on my old MacMini 32G using Ollama (remember to increase the context size!) and pi coding harness. I am getting good results with muse-glimmer running locally, with the caveat that everything runs slowly (e.g., give it a ta", "context": "runs on an old MacMini 32G", "use_case": "local coding model on Mac mini", "switched_from": "", "story": "Meta Muse Glimmer \u2013 open weights 30B local coding model"}]}