{"slug": "ollama", "name": "Ollama", "category": "ai", "category_name": "AI models and assistants", "description": "Tool for downloading and running open-weight large language models locally, with an optional cloud plan.", "recommenders": 38, "commenters": 123, "warners": 16, "score": 36.758, "rising": -1.82, "first": "2025-11-01", "latest": "2026-10-23", "why": "Liked as an easy way to run open models like Qwen and Gemma locally and privately, and for a $20/month Ollama Cloud plan that users find generous.", "why_is_quote": false, "has_verdict": true, "sentiment": {"recommend": 39, "neutral": 66, "compare": 9, "warn": 16}, "spark": [5, 1, 2, 2, 2, 2, 2, 11, 4, 2, 6, 0], "listed": true, "loved": ["Cloud plan value", "Local models"], "caveat": "Attribution concerns", "short_quote": "Download ollama and give it a try. It doesn't have to be that one but it's an easy on-ramp.", "icon": true, "domain": "ollama.com", "github_repo": null, "official_url": "https://ollama.com", "pricing": "freemium", "aliases": ["Ollama", "Ollama Cloud", "ollama"], "verdict": [{"text": "Liked as an easy way to run open models like Qwen and Gemma locally and privately,", "cites": [{"id": 782357552, "item_id": 48897347, "url": "https://news.ycombinator.com/item?id=48897347", "by": "andrew_ocs", "date": "2026-07-13", "stance": "recommend", "quote": "the most convenient way to run local LLMs", "excerpt": "hi, you should try https://ollama.com/ which is imo the most convenient way to run local LLMs", "context": "", "use_case": "run local LLMs", "switched_from": "", "story": "Ask HN: How do you use LLMs for private discussions?"}, {"id": 792981184, "item_id": 49561324, "url": "https://news.ycombinator.com/item?id=49561324", "by": "nucleative", "date": "2026-09-04", "stance": "recommend", "quote": "Download ollama and give it a try. It doesn't have to be that one but it's an easy on-ramp.", "excerpt": "1. Download ollama and give it a try. It doesn't have to be that one but it's an easy on-ramp. Just get a feel for what open source/open weights is capable of 2. Mention it whenever it comes up. Most people have no clue this is a thing and I think it's useful to make people aware. We need access to uncensored / unbiased LLM models - information wants to be free but there are plenty of businesses gunning for regulator", "context": "", "use_case": "running open-weights models locally", "switched_from": "", "story": "Which tools do Claude, Codex and Cursor choose? We measured 17k runs to find out"}, {"id": 764630288, "item_id": 47789393, "url": "https://news.ycombinator.com/item?id=47789393", "by": "0xbadcafebee", "date": "2026-04-15", "stance": "recommend", "quote": "Ollama is about 1000x easier to use.", "excerpt": "No mention of the fact that Ollama is about 1000x easier to use. Llama.cpp is a great project, but it's also one of the least user friendly pieces of software I've used. I don't think anyone in the project cares about normal users. I started with Ollama, and it was great. But I moved to llama.cpp to have more up-to-date fixes. I still use Ollama to pull and list my models because it's so easy. I then built my own set", "context": "", "use_case": "pulling and listing models", "switched_from": "", "story": "Stop Using Ollama"}, {"id": 777282688, "item_id": 48580168, "url": "https://news.ycombinator.com/item?id=48580168", "by": "pavan1989", "date": "2026-06-17", "stance": "recommend", "quote": "I used ollama with Qwen models works awesome", "excerpt": "I used ollama with Qwen models works awesome", "context": "", "use_case": "local coding models", "switched_from": "", "story": "Ask HN: Has anyone replaced Claude/GPT with a local model for daily coding?"}, {"id": 749553313, "item_id": 46847082, "url": "https://news.ycombinator.com/item?id=46847082", "by": "taravision", "date": "2026-02-01", "stance": "recommend", "quote": "It runs entirely on your machine via Ollama (I recommend gemma3:27b)", "excerpt": "Hi HN, I built taracode because I wanted an AI coding assistant for DevOps work that doesn't send my infrastructure code, credentials, or prompts to third parties. It runs entirely on your machine via Ollama (I recommend gemma3:27b) and has 58 built-in tools for the stuff I deal with daily: - Kubernetes: kubectl, helm - Terraform: plan, apply, state management - Docker: build, compose, logs - Cloud CLIs: AWS, Azure, ", "context": "", "use_case": "running local models", "switched_from": "", "story": "Show HN: Taracode \u2013 Open-source DevOps AI assistant that runs 100% locally"}]}, {"text": " and for a $20/month Ollama Cloud plan that users find generous.", "cites": [{"id": 778082928, "item_id": 48630183, "url": "https://news.ycombinator.com/item?id=48630183", "by": "Sanzig", "date": "2026-06-22", "stance": "recommend", "quote": "The $20/mo tier is very generous, in my experience.", "excerpt": "Take a look at Ollama Cloud: https://ollama.com/pricing You get access to a whole bunch of bleeding edge open models including GLM-5.2, Kimi K2.7, DeepSeek 4 Pro, etc. Inference is run on US/SG/EU cloud providers with zero data retention policies. The $20/mo tier is very generous, in my experience.", "context": "", "use_case": "hosted open models for coding", "switched_from": "", "story": "GLM 5.2 vs. Opus"}, {"id": 784790689, "item_id": 49049418, "url": "https://news.ycombinator.com/item?id=49049418", "by": "rglullis", "date": "2026-07-25", "stance": "recommend", "quote": "I never hit quota limits.", "excerpt": "I am using GLM-5.2 via Ollama Cloud in the $20/month plan. With the same plan I can get many different API keys that I use to run my OpenWebUI server, my opencode and pi dev sessions. I am usually running 2 to 4 sessions concurrently, and I never hit quota limits. At work I get Claude, and I was getting reports that I was spending $75 per hour of work on Opus.", "context": "", "use_case": "hosted open models", "switched_from": "", "story": "Open-weight AI is having its Kubernetes moment. Let's not ruin it"}, {"id": 777117794, "item_id": 48569862, "url": "https://news.ycombinator.com/item?id=48569862", "by": "pjerem", "date": "2026-06-17", "stance": "recommend", "quote": "I get so much usage out of it on a $20 Ollama cloud plan.", "excerpt": "I really like DeepSeek V4 Pro. It's pretty smart and I get so much usage out of it on a $20 Ollama cloud plan. With a good harness, that's my favorite model for any personal project. I use Opus 4.8 at work because i don't have to pay for it and of course I love it, but DeepSeek is like 80% there for one tenth of the price.", "context": "", "use_case": "cloud model hosting", "switched_from": "", "story": "GLM-5.2 is the new leading open weights model on Artificial Analysis"}, {"id": 770333329, "item_id": 48145833, "url": "https://news.ycombinator.com/item?id=48145833", "by": "pjerem", "date": "2026-05-15", "stance": "recommend", "quote": "Never reached the 5 hours usage limits of the $20 plan (on side projects)", "excerpt": "Hum, I'm using it [0] with my Ollama Cloud subscription since the last two weeks and I love it. Never reached the 5 hours usage limits of the $20 plan (on side projects) where I would reach it sometimes in ONE prompt with Opus. [0]: https://ollama.com/library/deepseek-v4-pro", "context": "", "use_case": "hosted model access", "switched_from": "", "story": "Cut Off: why access to frontier AI will soon be scarce and selective"}]}, {"text": " The most common complaint is that it takes llama.cpp's work without proper attribution,", "cites": [{"id": 793011680, "item_id": 49563230, "url": "https://news.ycombinator.com/item?id=49563230", "by": "KronisLV", "date": "2026-09-04", "stance": "warn", "quote": "Weren't people pissed off at them for a variety of reasons? Not giving attribution etc.", "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": 774444145, "item_id": 48402759, "url": "https://news.ycombinator.com/item?id=48402759", "by": "dwroberts", "date": "2026-06-04", "stance": "warn", "quote": "Highly recommend just dropping Ollama.", "excerpt": "Highly recommend just dropping Ollama. You can download binary releases of llama.cpp for every platform and run them trivially in 5 seconds. Ollama serves no purpose other than to take open source work and rebadge as its own, while providing inferior functionality", "context": "", "use_case": "running local models", "switched_from": "", "story": "Gemma 4 12B: A unified, encoder-free multimodal model"}, {"id": 766873600, "item_id": 47929600, "url": "https://news.ycombinator.com/item?id=47929600", "by": "anandkrshnn", "date": "2026-04-27", "stance": "warn", "quote": "The fact that I can\u2019t easily point another tool at the same GGUF file is a dealbreaker for me.", "excerpt": "Thanks for laying out the timeline so clearly. I\u2019ve been using Ollama for months because it \u2018just worked,\u2019 but I had no idea about the missing attribution or the hashed blob storage lock-in. The fact that I can\u2019t easily point another tool at the same GGUF file is a dealbreaker for me. Going to try llama-server today.", "context": "", "use_case": "running local models", "switched_from": "", "story": "Stop Using Ollama"}, {"id": 764638528, "item_id": 47789908, "url": "https://news.ycombinator.com/item?id=47789908", "by": "thot_experiment", "date": "2026-04-16", "stance": "warn", "quote": "Ollama is a vampire on the culture and their demise cannot come soon enough.", "excerpt": "I was pretty big on ollama, it seemed like a great default solution. I had alpha that it was a trash organization but I didn't listen because I just liked having a reliable inference backend that didn't require me to install torch. I switched to llama.cpp for everything maybe 6 months ago because of how fucking frustrating every one of my interactions with ollama (the organization) were. I wanna publicly apologize to", "context": "", "use_case": "local model inference", "switched_from": "", "story": "Stop Using Ollama"}]}, {"text": " and many say it lags llama.cpp on speed and features.", "cites": [{"id": 783500928, "item_id": 48968808, "url": "https://news.ycombinator.com/item?id=48968808", "by": "ollamer", "date": "2026-07-19", "stance": "warn", "quote": "on this VC funded joke it would be simply unusable", "excerpt": "https://github.com/ollama/ollama/issues/11772 A year and still no implementation for such a basic need as offloading MoE layers onto the CPU selectively. On llama.cpp I can get models like Qwen 35BA3B running partially on gpu/cpu with 40t/s on a laptop thanks to --n-cpu-moe but on this VC funded joke it would be simply unusable. I can't quite understand how you make a wrapper so much worse than the code you're rippin", "context": "", "use_case": "running local models", "switched_from": "", "story": "Ollama: All Aboard Open Models"}, {"id": 776947537, "item_id": 48559221, "url": "https://news.ycombinator.com/item?id=48559221", "by": "spockz", "date": "2026-06-16", "stance": "warn", "quote": "found a lot of performance improvement over ollama", "excerpt": "For what it is worth, I\u2019m on a similar machine. (9070XT,5900X) and found a lot of performance improvement over ollama by compiling llama.cpp and running with \u2014no-mmap and \u2014perf. The context is still quite small though. With online models I use contexts of at least 200k which is useful for longer running/more complicated commands. Locally I haven\u2019t gone much further than 8k. That is sufficient for small changes on sma", "context": "", "use_case": "local LLM inference", "switched_from": "", "story": "Running local models is good now"}, {"id": 791773009, "item_id": 49485813, "url": "https://news.ycombinator.com/item?id=49485813", "by": "wolvoleo", "date": "2026-08-29", "stance": "warn", "quote": "Ollama sucks, they dropped ROCm support for my card out of the blue", "excerpt": "I use Llama.cpp and OpenAI-style API of course (who doesn't, unless you use Ollama perhaps?) Ollama sucks, they dropped ROCm support for my card out of the blue, and the Vulkan replacement wasn't ready. The developer just shrugged and didn't care about the people affected so I moved to llama-server. Also, here's some other reasons to avoid ollama: https://sleepingrobots.com/dreams/stop-using-ollama/ And really, ollam", "context": "runs Linux server with multiple GPUs", "use_case": "running local LLMs", "switched_from": "", "story": "Run Qwen3.8 27B locally: real numbers from my Mac Studio"}, {"id": 762055826, "item_id": 47628489, "url": "https://news.ycombinator.com/item?id=47628489", "by": "bwannasek", "date": "2026-04-03", "stance": "warn", "quote": "due to some active bugs in ollama related to reasoning and streaming", "excerpt": "Using Gemma 4 with OpenCode was more challenging than expected due to some active bugs in ollama related to reasoning and streaming - I did a quick writeup in how I used llama.cpp instead of ollama and how to set it up to support multi-turn tool calls properly in case this is helpful to others: https://bernhardwannasek.com/using-gemma-4-for-agentic-coding/", "context": "", "use_case": "running local models", "switched_from": "", "story": "Google releases Gemma 4 open models"}]}], "praise": [{"label": "Cloud plan value", "summary": "Several users say the $20/month Ollama Cloud plan gives access to large open models and they rarely hit usage limits.", "quotes": [{"id": 778082928, "item_id": 48630183, "url": "https://news.ycombinator.com/item?id=48630183", "by": "Sanzig", "date": "2026-06-22", "stance": "recommend", "quote": "The $20/mo tier is very generous, in my experience.", "excerpt": "Take a look at Ollama Cloud: https://ollama.com/pricing You get access to a whole bunch of bleeding edge open models including GLM-5.2, Kimi K2.7, DeepSeek 4 Pro, etc. Inference is run on US/SG/EU cloud providers with zero data retention policies. The $20/mo tier is very generous, in my experience.", "context": "", "use_case": "hosted open models for coding", "switched_from": "", "story": "GLM 5.2 vs. Opus"}, {"id": 784790689, "item_id": 49049418, "url": "https://news.ycombinator.com/item?id=49049418", "by": "rglullis", "date": "2026-07-25", "stance": "recommend", "quote": "I never hit quota limits.", "excerpt": "I am using GLM-5.2 via Ollama Cloud in the $20/month plan. With the same plan I can get many different API keys that I use to run my OpenWebUI server, my opencode and pi dev sessions. I am usually running 2 to 4 sessions concurrently, and I never hit quota limits. At work I get Claude, and I was getting reports that I was spending $75 per hour of work on Opus.", "context": "", "use_case": "hosted open models", "switched_from": "", "story": "Open-weight AI is having its Kubernetes moment. Let's not ruin it"}, {"id": 770333329, "item_id": 48145833, "url": "https://news.ycombinator.com/item?id=48145833", "by": "pjerem", "date": "2026-05-15", "stance": "recommend", "quote": "Never reached the 5 hours usage limits of the $20 plan (on side projects)", "excerpt": "Hum, I'm using it [0] with my Ollama Cloud subscription since the last two weeks and I love it. Never reached the 5 hours usage limits of the $20 plan (on side projects) where I would reach it sometimes in ONE prompt with Opus. [0]: https://ollama.com/library/deepseek-v4-pro", "context": "", "use_case": "hosted model access", "switched_from": "", "story": "Cut Off: why access to frontier AI will soon be scarce and selective"}]}, {"label": "Local models", "summary": "Users report running Qwen, Gemma and GPT-OSS models well on their own GPUs and Macs.", "quotes": [{"id": 776634784, "item_id": 48539674, "url": "https://news.ycombinator.com/item?id=48539674", "by": "Peterz_shu", "date": "2026-06-15", "stance": "recommend", "quote": "Runs pretty well with Ollama on the Qwen models.", "excerpt": "I have a Mac mini (the square one, idk the model) with M3 chip. Runs pretty well with Ollama on the Qwen models. It seems like Qwen has done a great job with speed.", "context": "", "use_case": "local LLM on Mac mini", "switched_from": "", "story": "Ask HN: What does your local LLM setup looks like?"}, {"id": 795379683, "item_id": 49711230, "url": "https://news.ycombinator.com/item?id=49711230", "by": "Zambyte", "date": "2026-09-15", "stance": "recommend", "quote": "ollama to run qwen 3.8 27b on my single 7900 XTX", "excerpt": "Up until the last couple of months, I have treated LLMs as a supercharged stackoverflow. I would ask it questions on how to do something in a general sense, and then adapt the answer to my use case. Now, my entire programming flow does not even include an editor. The tools I use are: pi.dev to write and implement openspec specifications, herdr to manage many pi instances, and ollama to run qwen 3.8 27b on my single 7", "context": "", "use_case": "running local models", "switched_from": "", "story": "How much of F-Droid is LLM generated?"}, {"id": 733134304, "item_id": 45820894, "url": "https://news.ycombinator.com/item?id=45820894", "by": "EagnaIonat", "date": "2025-11-05", "stance": "recommend", "quote": "For my local work I use Ollama. (M4 Max 128GB)", "excerpt": "The more recent LLMs work fine on an M1 mac. Can't speak for Windows/Linux. There was even a recent release of Granite4 that runs on a Raspberry Pi. https://github.com/Jewelzufo/granitepi-4-nano For my local work I use Ollama. (M4 Max 128GB) - gpt-oss. 20b or 120b depending on complexity of use cases. - granite4 for speed and lower complexity (around the same as gpt20b).", "context": "M4 Max 128GB Mac user", "use_case": "running local LLMs", "switched_from": "", "story": "Developers are choosing older AI models"}]}, {"label": "Easy setup", "summary": "Often described as an easy on-ramp that just works, much simpler to use than llama.cpp.", "quotes": [{"id": 764630288, "item_id": 47789393, "url": "https://news.ycombinator.com/item?id=47789393", "by": "0xbadcafebee", "date": "2026-04-15", "stance": "recommend", "quote": "Ollama is about 1000x easier to use.", "excerpt": "No mention of the fact that Ollama is about 1000x easier to use. Llama.cpp is a great project, but it's also one of the least user friendly pieces of software I've used. I don't think anyone in the project cares about normal users. I started with Ollama, and it was great. But I moved to llama.cpp to have more up-to-date fixes. I still use Ollama to pull and list my models because it's so easy. I then built my own set", "context": "", "use_case": "pulling and listing models", "switched_from": "", "story": "Stop Using Ollama"}, {"id": 792981184, "item_id": 49561324, "url": "https://news.ycombinator.com/item?id=49561324", "by": "nucleative", "date": "2026-09-04", "stance": "recommend", "quote": "Download ollama and give it a try. It doesn't have to be that one but it's an easy on-ramp.", "excerpt": "1. Download ollama and give it a try. It doesn't have to be that one but it's an easy on-ramp. Just get a feel for what open source/open weights is capable of 2. Mention it whenever it comes up. Most people have no clue this is a thing and I think it's useful to make people aware. We need access to uncensored / unbiased LLM models - information wants to be free but there are plenty of businesses gunning for regulator", "context": "", "use_case": "running open-weights models locally", "switched_from": "", "story": "Which tools do Claude, Codex and Cursor choose? We measured 17k runs to find out"}, {"id": 788363984, "item_id": 49272749, "url": "https://news.ycombinator.com/item?id=49272749", "by": "greenmoonx", "date": "2026-08-12", "stance": "recommend", "quote": "I use Ollama (CLI and their npm as a dev) + HF because it just works.", "excerpt": "The whole AI \"industry\" needs to take a course in digital marketing. Then spend about 5-10 years minimum building web apps to get an idea how releasing a build and distribution works. Very unimpressed by how everything in \"AI\" is brought to market. It's like Apple Watch fanboys but with Linux style evangelism. Nobody wants that and it doesn't make us think you are \"smart\" and we're \"dumb\" if your tech is not easy to ", "context": "developer", "use_case": "running local LLMs as a dev", "switched_from": "", "story": "llama.cpp"}]}, {"label": "Privacy", "summary": "Used to keep prompts and data on your own machine, with nothing sent to third parties.", "quotes": [{"id": 749553313, "item_id": 46847082, "url": "https://news.ycombinator.com/item?id=46847082", "by": "taravision", "date": "2026-02-01", "stance": "recommend", "quote": "It runs entirely on your machine via Ollama (I recommend gemma3:27b)", "excerpt": "Hi HN, I built taracode because I wanted an AI coding assistant for DevOps work that doesn't send my infrastructure code, credentials, or prompts to third parties. It runs entirely on your machine via Ollama (I recommend gemma3:27b) and has 58 built-in tools for the stuff I deal with daily: - Kubernetes: kubectl, helm - Terraform: plan, apply, state management - Docker: build, compose, logs - Cloud CLIs: AWS, Azure, ", "context": "", "use_case": "running local models", "switched_from": "", "story": "Show HN: Taracode \u2013 Open-source DevOps AI assistant that runs 100% locally"}, {"id": 735868145, "item_id": 45991759, "url": "https://news.ycombinator.com/item?id=45991759", "by": "vyntral", "date": "2025-11-20", "stance": "recommend", "quote": "using Ollama (runs 100% locally)", "excerpt": "Hi HN! I built God's Eye to solve a problem I had doing bug bounties: existing recon tools found subdomains, but I still had to manually analyze them for security issues. God's Eye combines subdomain enumeration with AI security analysis using Ollama (runs 100% locally): \u2022 11 passive sources + DNS brute-forcing \u2022 HTTP probing + tech detection \u2022 Real-time CVE detection via function calling \u2022 JavaScript security review", "context": "bug bounty hunter", "use_case": "local LLM for security analysis", "switched_from": "", "story": "Show HN: God's Eye \u2013 AI-powered subdomain recon with local LLM"}, {"id": 732478096, "item_id": 45779881, "url": "https://news.ycombinator.com/item?id=45779881", "by": "Greenpants", "date": "2025-11-01", "stance": "recommend", "quote": "It runs Ollama + Open WebUI with GPT-OSS 120b as well as GLM4.5-Air", "excerpt": "I got a personal Mac Studio M4 Max with 128GB RAM for a silent, relatively power-efficient yet powerful home server. It runs Ollama + Open WebUI with GPT-OSS 120b as well as GLM4.5-Air (default quantisations). I rarely ever use ChatGPT anymore. Love that all data stays at home. I connect remotely only via VPN (my phone enables this automatically via Tasker). I'm 50% brainstorming ideas with it, asking critical questi", "context": "Mac Studio M4 Max home server", "use_case": "home LLM server", "switched_from": "", "story": "Ask HN: Who uses open LLMs and coding assistants locally? Share setup and laptop"}]}, {"label": "Developer API", "summary": "Developers build tools on top of its API and npm package.", "quotes": [{"id": 744225152, "item_id": 46514072, "url": "https://news.ycombinator.com/item?id=46514072", "by": "billylo", "date": "2026-01-06", "stance": "recommend", "quote": "For consistent LLM behaviour, you can use ollama api with your model of choice to generate.", "excerpt": "Windows and macOS does come with a small model for generating text completion. You can write a wrapper for your own TUI to access them platform agnostically. For consistent LLM behaviour, you can use ollama api with your model of choice to generate. https://docs.ollama.com/api/generate Chrome has a built-in Gemini Nano too. But there isn't an official way to use it outside chrome yet.", "context": "", "use_case": "consistent local text completion", "switched_from": "", "story": "Ask HN: What's a standard way for apps to request text completion as a service?"}, {"id": 788363984, "item_id": 49272749, "url": "https://news.ycombinator.com/item?id=49272749", "by": "greenmoonx", "date": "2026-08-12", "stance": "recommend", "quote": "I use Ollama (CLI and their npm as a dev) + HF because it just works.", "excerpt": "The whole AI \"industry\" needs to take a course in digital marketing. Then spend about 5-10 years minimum building web apps to get an idea how releasing a build and distribution works. Very unimpressed by how everything in \"AI\" is brought to market. It's like Apple Watch fanboys but with Linux style evangelism. Nobody wants that and it doesn't make us think you are \"smart\" and we're \"dumb\" if your tech is not easy to ", "context": "developer", "use_case": "running local LLMs as a dev", "switched_from": "", "story": "llama.cpp"}, {"id": 735038464, "item_id": 45939904, "url": "https://news.ycombinator.com/item?id=45939904", "by": "mfyz", "date": "2025-11-15", "stance": "recommend", "quote": "Summaries via OpenAI or local models (I use Gemma 4B on Ollama)", "excerpt": "I built a small open-source tool to deal with my own content overload. It monitors newsletters, blogs, and RSS feeds, extracts new posts, summarizes them using LLMs (OpenAI or local Ollama), and sends one Slack digest each morning. Everything runs in a single Docker container using Node.js + SQLite. \u2022 RSS, CSS-selector, or LLM-based extraction \u2022 Summaries via OpenAI or local models (I use Gemma 4B on Ollama) \u2022 Simple", "context": "", "use_case": "local LLM summaries", "switched_from": "", "story": "I built an OSS newsletter digester that uses AI to send me daily Slack summaries"}]}], "caveats": [{"label": "Attribution concerns", "summary": "Critics say it rebadges llama.cpp work without giving proper credit and recommend alternatives.", "quotes": [{"id": 774444145, "item_id": 48402759, "url": "https://news.ycombinator.com/item?id=48402759", "by": "dwroberts", "date": "2026-06-04", "stance": "warn", "quote": "Highly recommend just dropping Ollama.", "excerpt": "Highly recommend just dropping Ollama. You can download binary releases of llama.cpp for every platform and run them trivially in 5 seconds. Ollama serves no purpose other than to take open source work and rebadge as its own, while providing inferior functionality", "context": "", "use_case": "running local models", "switched_from": "", "story": "Gemma 4 12B: A unified, encoder-free multimodal model"}, {"id": 793011680, "item_id": 49563230, "url": "https://news.ycombinator.com/item?id=49563230", "by": "KronisLV", "date": "2026-09-04", "stance": "warn", "quote": "Weren't people pissed off at them for a variety of reasons? Not giving attribution etc.", "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": 764638528, "item_id": 47789908, "url": "https://news.ycombinator.com/item?id=47789908", "by": "thot_experiment", "date": "2026-04-16", "stance": "warn", "quote": "Ollama is a vampire on the culture and their demise cannot come soon enough.", "excerpt": "I was pretty big on ollama, it seemed like a great default solution. I had alpha that it was a trash organization but I didn't listen because I just liked having a reliable inference backend that didn't require me to install torch. I switched to llama.cpp for everything maybe 6 months ago because of how fucking frustrating every one of my interactions with ollama (the organization) were. I wanna publicly apologize to", "context": "", "use_case": "local model inference", "switched_from": "", "story": "Stop Using Ollama"}]}, {"label": "Lags llama.cpp", "summary": "Users report better performance and missing features, such as MoE CPU offloading, when using llama.cpp directly.", "quotes": [{"id": 783500928, "item_id": 48968808, "url": "https://news.ycombinator.com/item?id=48968808", "by": "ollamer", "date": "2026-07-19", "stance": "warn", "quote": "on this VC funded joke it would be simply unusable", "excerpt": "https://github.com/ollama/ollama/issues/11772 A year and still no implementation for such a basic need as offloading MoE layers onto the CPU selectively. On llama.cpp I can get models like Qwen 35BA3B running partially on gpu/cpu with 40t/s on a laptop thanks to --n-cpu-moe but on this VC funded joke it would be simply unusable. I can't quite understand how you make a wrapper so much worse than the code you're rippin", "context": "", "use_case": "running local models", "switched_from": "", "story": "Ollama: All Aboard Open Models"}, {"id": 776947537, "item_id": 48559221, "url": "https://news.ycombinator.com/item?id=48559221", "by": "spockz", "date": "2026-06-16", "stance": "warn", "quote": "found a lot of performance improvement over ollama", "excerpt": "For what it is worth, I\u2019m on a similar machine. (9070XT,5900X) and found a lot of performance improvement over ollama by compiling llama.cpp and running with \u2014no-mmap and \u2014perf. The context is still quite small though. With online models I use contexts of at least 200k which is useful for longer running/more complicated commands. Locally I haven\u2019t gone much further than 8k. That is sufficient for small changes on sma", "context": "", "use_case": "local LLM inference", "switched_from": "", "story": "Running local models is good now"}, {"id": 791773009, "item_id": 49485813, "url": "https://news.ycombinator.com/item?id=49485813", "by": "wolvoleo", "date": "2026-08-29", "stance": "warn", "quote": "Ollama sucks, they dropped ROCm support for my card out of the blue", "excerpt": "I use Llama.cpp and OpenAI-style API of course (who doesn't, unless you use Ollama perhaps?) Ollama sucks, they dropped ROCm support for my card out of the blue, and the Vulkan replacement wasn't ready. The developer just shrugged and didn't care about the people affected so I moved to llama-server. Also, here's some other reasons to avoid ollama: https://sleepingrobots.com/dreams/stop-using-ollama/ And really, ollam", "context": "runs Linux server with multiple GPUs", "use_case": "running local LLMs", "switched_from": "", "story": "Run Qwen3.8 27B locally: real numbers from my Mac Studio"}]}, {"label": "Bugs", "summary": "Users report bugs with reasoning and streaming, and a prompt injection issue that only Ollama's developers can fix.", "quotes": [{"id": 762055826, "item_id": 47628489, "url": "https://news.ycombinator.com/item?id=47628489", "by": "bwannasek", "date": "2026-04-03", "stance": "warn", "quote": "due to some active bugs in ollama related to reasoning and streaming", "excerpt": "Using Gemma 4 with OpenCode was more challenging than expected due to some active bugs in ollama related to reasoning and streaming - I did a quick writeup in how I used llama.cpp instead of ollama and how to set it up to support multi-turn tool calls properly in case this is helpful to others: https://bernhardwannasek.com/using-gemma-4-for-agentic-coding/", "context": "", "use_case": "running local models", "switched_from": "", "story": "Google releases Gemma 4 open models"}, {"id": 787261120, "item_id": 49203820, "url": "https://news.ycombinator.com/item?id=49203820", "by": "pavelai", "date": "2026-08-06", "stance": "warn", "quote": "for tools like Ollama there is no solution except of to fix it by Ollama developers", "excerpt": "Prompt injection allows third-party to inject a system prompt with simple message, by inserting special HTML-like sequence (details below). Some of the issues are well-known and pretty old (almost 2 years for Transformers library) Issues: \u2022 Ollama: https://github.com/ollama/ollama/issues/15931 \u2022 HuggingFace: https://github.com/huggingface/transformers/issues/29279 (labeled as feature request) https://github.com/huggi", "context": "", "use_case": "running local models", "switched_from": "", "story": "Prompt Injection Vulnerability in Ollama, Gemma4 and HuggingFace's Transformers"}]}, {"label": "Small default context", "summary": "It defaults to a very small context window, which can break agent harnesses.", "quotes": [{"id": 798924708, "item_id": 49932794, "url": "https://news.ycombinator.com/item?id=49932794", "by": "nfriedly", "date": "2026-10-02", "stance": "warn", "quote": "ollama defaults to incredibly tiny amounts of context", "excerpt": "I really like Pi, especially for when you're running a local model and your context is particularly limited. But even when running a big model, I think building up your environment to be exactly what you want can be really powerful. When I was first getting started and didn't realize that ollama defaults to incredibly tiny amounts of context, Pi was the only harness that actually worked because the rest of them used ", "context": "", "use_case": "running local models", "switched_from": "", "story": "Pi 1.0"}]}, {"label": "Model storage lock-in", "summary": "Models are stored as hashed blobs, so other tools cannot easily reuse the same GGUF files.", "quotes": [{"id": 766873600, "item_id": 47929600, "url": "https://news.ycombinator.com/item?id=47929600", "by": "anandkrshnn", "date": "2026-04-27", "stance": "warn", "quote": "The fact that I can\u2019t easily point another tool at the same GGUF file is a dealbreaker for me.", "excerpt": "Thanks for laying out the timeline so clearly. I\u2019ve been using Ollama for months because it \u2018just worked,\u2019 but I had no idea about the missing attribution or the hashed blob storage lock-in. The fact that I can\u2019t easily point another tool at the same GGUF file is a dealbreaker for me. Going to try llama-server today.", "context": "", "use_case": "running local models", "switched_from": "", "story": "Stop Using Ollama"}]}, {"label": "Monetization worries", "summary": "Some users left after the team started monetizing, and others call it a VC-funded push toward its cloud platform.", "quotes": [{"id": 732896433, "item_id": 45806027, "url": "https://news.ycombinator.com/item?id=45806027", "by": "stuxnet79", "date": "2025-11-03", "stance": "warn", "quote": "I got spooked when the Ollama team started monetizing", "excerpt": "It seems like the ecosystem around these tools has matured quite rapidly. I am somewhat familiar with Open WebUI, however, the last time I played around with it, I got the sense that it was merely a front-end to Ollama, the llm command line tool & it didn't have any capabilities outside of that. I got spooked when the Ollama team started monetizing so I ended up doing more research into llama.cpp and realized it coul", "context": "", "use_case": "running local LLMs", "switched_from": "", "story": "Ask HN: Who uses open LLMs and coding assistants locally? Share setup and laptop"}, {"id": 791773009, "item_id": 49485813, "url": "https://news.ycombinator.com/item?id=49485813", "by": "wolvoleo", "date": "2026-08-29", "stance": "warn", "quote": "Ollama sucks, they dropped ROCm support for my card out of the blue", "excerpt": "I use Llama.cpp and OpenAI-style API of course (who doesn't, unless you use Ollama perhaps?) Ollama sucks, they dropped ROCm support for my card out of the blue, and the Vulkan replacement wasn't ready. The developer just shrugged and didn't care about the people affected so I moved to llama-server. Also, here's some other reasons to avoid ollama: https://sleepingrobots.com/dreams/stop-using-ollama/ And really, ollam", "context": "runs Linux server with multiple GPUs", "use_case": "running local LLMs", "switched_from": "", "story": "Run Qwen3.8 27B locally: real numbers from my Mac Studio"}, {"id": 783500928, "item_id": 48968808, "url": "https://news.ycombinator.com/item?id=48968808", "by": "ollamer", "date": "2026-07-19", "stance": "warn", "quote": "on this VC funded joke it would be simply unusable", "excerpt": "https://github.com/ollama/ollama/issues/11772 A year and still no implementation for such a basic need as offloading MoE layers onto the CPU selectively. On llama.cpp I can get models like Qwen 35BA3B running partially on gpu/cpu with 40t/s on a laptop thanks to --n-cpu-moe but on this VC funded joke it would be simply unusable. I can't quite understand how you make a wrapper so much worse than the code you're rippin", "context": "", "use_case": "running local models", "switched_from": "", "story": "Ollama: All Aboard Open Models"}]}], "who": [{"by": "ToriTech", "context": "learning AI security", "url": "https://news.ycombinator.com/item?id=49721567", "date": "2026-09-16"}, {"by": "jabrotapu", "context": "freelancer building on-prem GPU clusters", "url": "https://news.ycombinator.com/item?id=49661006", "date": "2026-09-11"}, {"by": "JacobJack", "context": "Windows user", "url": "https://news.ycombinator.com/item?id=49404004", "date": "2026-08-22"}, {"by": "nater5000", "context": "runs a home server with an RTX 3090 Ti", "url": "https://news.ycombinator.com/item?id=49391156", "date": "2026-08-21"}, {"by": "greenmoonx", "context": "developer", "url": "https://news.ycombinator.com/item?id=49272749", "date": "2026-08-12"}, {"by": "kentonv", "context": "works on Cloudflare OS", "url": "https://news.ycombinator.com/item?id=49189692", "date": "2026-08-05"}, {"by": "Shorel", "context": "work pays for cloud subscription", "url": "https://news.ycombinator.com/item?id=48737204", "date": "2026-06-30"}, {"by": "trollbridge", "context": "leads a team of 3 developers, 1 designer, 1 non-technical user", "url": "https://news.ycombinator.com/item?id=48503533", "date": "2026-06-12"}, {"by": "bastawhiz", "context": "Mac Studio owner", "url": "https://news.ycombinator.com/item?id=48168433", "date": "2026-05-17"}, {"by": "raxxorraxor", "context": "developer", "url": "https://news.ycombinator.com/item?id=47396204", "date": "2026-03-16"}, {"by": "rwyinuse", "context": "runs a home server", "url": "https://news.ycombinator.com/item?id=47148626", "date": "2026-02-24"}, {"by": "jakeasmith", "context": "home lab user", "url": "https://news.ycombinator.com/item?id=46681531", "date": "2026-01-19"}], "tags": ["for running models locally", "for privacy-sensitive work", "for Mac users", "for developers", "for beginners"], "use_cases": ["running local models", "running local llms", "local models", "local inference", "local llm inference", "serving local models"], "compared": [{"slug": "llama-cpp", "name": "llama.cpp", "co": 31, "switched_from": 0, "switched_to": 14}, {"slug": "gemma", "name": "Gemma", "co": 15, "switched_from": 0, "switched_to": 0}, {"slug": "lm-studio", "name": "LM Studio", "co": 12, "switched_from": 0, "switched_to": 1}, {"slug": "claude", "name": "Claude", "co": 12, "switched_from": 0, "switched_to": 0}, {"slug": "opencode", "name": "OpenCode", "co": 12, "switched_from": 0, "switched_to": 0}, {"slug": "claude-code", "name": "Claude Code", "co": 11, "switched_from": 0, "switched_to": 0}, {"slug": "qwen", "name": "Qwen", "co": 10, "switched_from": 0, "switched_to": 0}, {"slug": "claude-opus", "name": "Claude Opus", "co": 9, "switched_from": 0, "switched_to": 0}], "makers": [{"id": 753379857, "item_id": 47086241, "url": "https://news.ycombinator.com/item?id=47086241", "by": "laramie_co", "date": "2026-02-20", "stance": "recommend", "quote": "Ollama for inference.", "excerpt": "Hey HN \u2014 maker here. Quick context on why I built this: I was personally spending ~$1,200/month on Claude API (I use it for everything \u2014 coding, writing, analysis). That\u2019s $14,400/year. Even if I dropped to a single $20/month sub, every query I send tells Anthropic what I\u2019m working on, what I\u2019m thinking about, what problems I have. I wanted Claude-quality inference without the cloud dependency. That\u2019s not possible ye", "context": "", "use_case": "local inference", "switched_from": "", "story": "Show HN: VaultAI \u2013 42 AI models on a portable SSD, works offline ($399)"}], "threads": [{"url": "https://news.ycombinator.com/item?id=45771870", "title": "Ask HN: Who uses open LLMs and coding assistants locally? Share setup and laptop", "mentions": 7}, {"url": "https://news.ycombinator.com/item?id=48542100", "title": "Ask HN: Has anyone replaced Claude/GPT with a local model for daily coding?", "mentions": 5}, {"url": "https://news.ycombinator.com/item?id=47788385", "title": "Stop Using Ollama", "mentions": 4}, {"url": "https://news.ycombinator.com/item?id=49686380", "title": "Ask HN: What are you working on? (September 2026)", "mentions": 3}, {"url": "https://news.ycombinator.com/item?id=49557206", "title": "Which tools do Claude, Codex and Cursor choose? We measured 17k runs to find out", "mentions": 3}, {"url": "https://news.ycombinator.com/item?id=47863217", "title": "Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model", "mentions": 3}], "sentiment_by_month": {"2025-11": {"recommend": 5, "neutral": 4, "warn": 1}, "2025-12": {"neutral": 2, "recommend": 1}, "2026-01": {"recommend": 2, "neutral": 2, "compare": 1}, "2026-02": {"recommend": 2, "warn": 1, "neutral": 1}, "2026-03": {"neutral": 5, "recommend": 2}, "2026-04": {"neutral": 6, "warn": 5, "recommend": 2, "compare": 1}, "2026-05": {"compare": 2, "neutral": 1, "recommend": 2}, "2026-06": {"neutral": 12, "warn": 3, "recommend": 11, "compare": 1}, "2026-07": {"neutral": 5, "recommend": 4, "warn": 1}, "2026-08": {"neutral": 13, "warn": 2, "compare": 1, "recommend": 2}, "2026-09": {"neutral": 8, "recommend": 6, "warn": 2, "compare": 3}, "2026-10": {"neutral": 7, "warn": 1}}, "quotes": [{"id": 803325169, "item_id": 50207823, "url": "https://news.ycombinator.com/item?id=50207823", "by": "jimmar", "date": "2026-10-23", "stance": "neutral", "quote": "I see the arrow on the Ollama icon telling me I have to relaunch it to apply updates.", "excerpt": "> Most software, once mature, is mostly done. The caveat being that most software never matures. Maybe it's a cynical take, but virtually all software I use regularly is updated at least once a month. Right now, I'm using Chrome, and I see the \"Relaunch to update\" button waiting for me to press it. I see the arrow on the Ollama icon telling me I have to relaunch it to apply updates. I have Visual Studio Code running ", "context": "", "use_case": "running local LLMs", "switched_from": "", "story": "Is this the new cyber normal?"}, {"id": 801752545, "item_id": 50109534, "url": "https://news.ycombinator.com/item?id=50109534", "by": "alexpotato", "date": "2026-10-15", "stance": "neutral", "quote": "Been using GLM 5.3 Flash via Ollama Cloud basic subscription", "excerpt": "Been using GLM 5.3 Flash via Ollama Cloud basic subscription: if this is the only model I had to use for the rest of my life, I would be cool with that. It's good at a wide variety of tasks including reverse engineering, design, working with ESP 32 etc. I think Muse Spark 1.3 is better but GLM 5.3 Flash is an excellent daily driver.", "context": "", "use_case": "hosting GLM model", "switched_from": "", "story": "Muse Spark Open"}, {"id": 801586577, "item_id": 50099161, "url": "https://news.ycombinator.com/item?id=50099161", "by": "dv35z", "date": "2026-10-15", "stance": "neutral", "quote": "I used OpenCode to help me install ollama, Qwen local LLM, and connect OpenCode to it.", "excerpt": "I used OpenCode to help me install ollama, Qwen local LLM, and connect OpenCode to it. The computer is very slow (old Dell Optiplex running Mint Debian Linux), but it worked! So cool to turn off the network, and still be able to chit-chat with Qwen model (\"Tell me a joke about...\").", "context": "", "use_case": "running local LLMs", "switched_from": "", "story": "OpenCode 2.0"}, {"id": 800326738, "item_id": 50020421, "url": "https://news.ycombinator.com/item?id=50020421", "by": "nico", "date": "2026-10-09", "stance": "neutral", "quote": "I experiment a lot with models I can run locally via ollama", "excerpt": "Hybrid. I use Claude/codex subscription for most coding tasks, but I experiment a lot with models I can run locally via ollama, and also been building my own local classifier models which run and train on CPU (they can classify text, but also play doom): https://github.com/nicobrenner/jeffy", "context": "", "use_case": "running local models", "switched_from": "", "story": "Ask HN: Are you a subscribed LLM or a locally deployed open-source model?"}, {"id": 800199233, "item_id": 50012452, "url": "https://news.ycombinator.com/item?id=50012452", "by": "pjerem", "date": "2026-10-08", "stance": "neutral", "quote": "I used GLM-5.3 as usual from Ollama cloud", "excerpt": "IDK what happened today but I used GLM-5.3 as usual from Ollama cloud and it was so fast it generated entire documents like instantly. The reasoning and the result document were done after less than 1 or 2 seconds. Have Ollama suddenly bought GPU capacity?", "context": "", "use_case": "hosted model inference", "switched_from": "", "story": "Why isn't the industry freaking out about DeepSeek 4.1 Flash?"}, {"id": 799012225, "item_id": 49938264, "url": "https://news.ycombinator.com/item?id=49938264", "by": "BrentPollard", "date": "2026-10-02", "stance": "neutral", "quote": "currently building with qwen3.5 via Ollama", "excerpt": "Location: Toronto, Ontario, Canada Remote: Yes Willing to relocate: No Technologies: SQL (especially Postgres), Python, AWS, AI (currently building with qwen3.5 via Ollama), SAS, C, C++, Java, Dart + Flutter, JavaScript/TypeScript, some GCP, learning Rust. Resume: https://drive.google.com/file/d/1dgXWJz4Cd57c3_xedJmrTOBqmiY1IKKU/view?usp=drivesdk GitHub: github.com/BProgramming Email: brentjpollard@gmail.com I build ", "context": "", "use_case": "running local models", "switched_from": "", "story": "Ask HN: Who wants to be hired? (October 2026)"}, {"id": 798924708, "item_id": 49932794, "url": "https://news.ycombinator.com/item?id=49932794", "by": "nfriedly", "date": "2026-10-02", "stance": "warn", "quote": "ollama defaults to incredibly tiny amounts of context", "excerpt": "I really like Pi, especially for when you're running a local model and your context is particularly limited. But even when running a big model, I think building up your environment to be exactly what you want can be really powerful. When I was first getting started and didn't realize that ollama defaults to incredibly tiny amounts of context, Pi was the only harness that actually worked because the rest of them used ", "context": "", "use_case": "running local models", "switched_from": "", "story": "Pi 1.0"}, {"id": 798727953, "item_id": 49920497, "url": "https://news.ycombinator.com/item?id=49920497", "by": "pratikdeoghare", "date": "2026-10-01", "stance": "neutral", "quote": "Here I am using it as frontend to Jupyter and ollama [2].", "excerpt": "I have created Brashtag language for this and other things [1]. Here I am using it as frontend to Jupyter and ollama [2]. [1] https://github.com/PratikDeoghare/brashtag [2] https://youtu.be/IMXgIE0Vljg?si=sBmmZ1nHlbHifKBd", "context": "", "use_case": "local LLM backend", "switched_from": "", "story": "Show HN: Ledge.sh \u2013 Runnable Markdown Notes"}, {"id": 797115345, "item_id": 49819709, "url": "https://news.ycombinator.com/item?id=49819709", "by": "ezulabs", "date": "2026-09-23", "stance": "recommend", "quote": "you shall use CC, serve it via ollama", "excerpt": "you shall use CC, serve it via ollama, you need to patch it as CC's timestamp ruins prefill cache.", "context": "", "use_case": "serving local models", "switched_from": "", "story": "Ask HN: Which harness do you use with Qwen3.8 27B? And why?"}, {"id": 797084817, "item_id": 49817801, "url": "https://news.ycombinator.com/item?id=49817801", "by": "everforward", "date": "2026-09-23", "stance": "recommend", "quote": "I pay for the $20/month Ollama plan", "excerpt": "I use Pi and mostly open weight models. I pay for the $20/month Ollama plan and use Deepseek and GLM through that. I\u2019ve never hit the limits on it, but I tend to ask for targeted things rather than \u201cimplement a whole feature in one prompt\u201d. I do keep an OpenRouter account topped up for things that Ollama doesn\u2019t have. 99% of my usage there is embeddings, the other 1% is wanting to test some new model Ollama doesn\u2019t h", "context": "", "use_case": "hosted open-weight models", "switched_from": "", "story": "Strands Harness"}, {"id": 796742080, "item_id": 49796380, "url": "https://news.ycombinator.com/item?id=49796380", "by": "sarfaraznaushad", "date": "2026-09-22", "stance": "recommend", "quote": "I'm always excited to run models on my local machine. I use Ollama for that.", "excerpt": "I'm always excited to run models on my local machine. I use Ollama for that.", "context": "", "use_case": "running local models", "switched_from": "", "story": "Fable 5 \u2013 Median thinking declined in August"}, {"id": 796649601, "item_id": 49790600, "url": "https://news.ycombinator.com/item?id=49790600", "by": "david-ndungu", "date": "2026-09-21", "stance": "compare", "quote": "Zerfoo ran Gemma 3 1B at 235 tok/s in one recorded benchmark versus 188 tok/s with Ollama 0.17.7", "excerpt": "I\u2019ve been tinkering with machine learning since around 2018. My original motivation was an autonomous trading system I was building in Go. At the time I used TensorFlow for the models and wrote a gRPC service so the Go trading system could talk to them. I didn\u2019t particularly like that boundary. I wanted the model to be part of the application rather than a separate Python service. I also wanted to understand what was", "context": "", "use_case": "LLM inference", "switched_from": "", "story": "Zerfoo \u2013 a pure Go framework for creating, training and running AI models"}, {"id": 796128081, "item_id": 49758005, "url": "https://news.ycombinator.com/item?id=49758005", "by": "HoldOnAMinute", "date": "2026-09-18", "stance": "neutral", "quote": "When I use Claude Code locally with Ollama/Quen", "excerpt": "No - Here is what I mean. When I use Claude Code locally with Ollama/Quen, it can do the work just fine - like creating a script - but it just outputs it, rather than placing it into a file. So far I can't convince it to actually WRITE to a file. With Claude Code, it just works. Recursing around the file system, it just does it. I can't get that working locally.", "context": "", "use_case": "running local models", "switched_from": "", "story": "Ask HN: What happens when Claude Code goes away?"}, {"id": 795545072, "item_id": 49721567, "url": "https://news.ycombinator.com/item?id=49721567", "by": "ToriTech", "date": "2026-09-16", "stance": "recommend", "quote": "Download Ollama (you can install the models themselves onto a USB drive so you don't fill up your hard drive with LLMs)", "excerpt": "Assuming you want a plan with a $0 budget, it's probably best if you start with TryHackMe and Portswigger Academy. The free tier on TryHackMe gives you access to a lot of labs and challenges, mostly focusing on Web security, forensics, and some blue team work, but they're starting to add more AI content as of late. Portswigger Academy is completely free, made by the people who created Burpsuite, and has a lot of high", "context": "learning AI security", "use_case": "local LLMs for security experiments", "switched_from": "", "story": "Started Learning Pentesting"}, {"id": 795379683, "item_id": 49711230, "url": "https://news.ycombinator.com/item?id=49711230", "by": "Zambyte", "date": "2026-09-15", "stance": "recommend", "quote": "ollama to run qwen 3.8 27b on my single 7900 XTX", "excerpt": "Up until the last couple of months, I have treated LLMs as a supercharged stackoverflow. I would ask it questions on how to do something in a general sense, and then adapt the answer to my use case. Now, my entire programming flow does not even include an editor. The tools I use are: pi.dev to write and implement openspec specifications, herdr to manage many pi instances, and ollama to run qwen 3.8 27b on my single 7", "context": "", "use_case": "running local models", "switched_from": "", "story": "How much of F-Droid is LLM generated?"}, {"id": 795281618, "item_id": 49705101, "url": "https://news.ycombinator.com/item?id=49705101", "by": "basetensucks", "date": "2026-09-14", "stance": "neutral", "quote": "using a local `ollama` model", "excerpt": "I've been working on `ankix` (https://github.com/joshgummersall/ankix), a CLI tool to turn YouTube/podcast transcripts, Kindle highlights, and other text into \"glossed\" (read: contextually translated) Anki cards using a local `ollama` model. I've also been working on `summond` (https://github.com/joshgummersall/summond), a developer/agent friendly way to interact with OSX Launch Control. It aims to support all releva", "context": "", "use_case": "local model for glossing", "switched_from": "", "story": "Ask HN: What are you working on? (September 2026)"}, {"id": 795259280, "item_id": 49703705, "url": "https://news.ycombinator.com/item?id=49703705", "by": "sbankowi", "date": "2026-09-14", "stance": "compare", "quote": "why should I not just use those directly if I am focused on local first models", "excerpt": "I am looked at the website, but I am not able to understand what problem you are trying to solve. Or why would I use you? (How are you differentiating yourself?) I see you are reliant on ollama and llama.cpp (for example), so why should I not just use those directly if I am focused on local first models. (llama.cpp already makes recommendations on what models to run.) If the argument is then that you are more of an a", "context": "", "use_case": "running local models", "switched_from": "", "story": "Show HN: I built Otis, a minimal AI agent that runs local models out of the box"}, {"id": 795092320, "item_id": 49693270, "url": "https://news.ycombinator.com/item?id=49693270", "by": "darkhorse_chenn", "date": "2026-09-14", "stance": "neutral", "quote": "why not we train a local ollama  model 1.5 b parameters", "excerpt": "i have been working on like a different project like since recently TALA has been opensourced , i analysed how it constructs flow using dag after installing it on my linux , , i thought like why not we train a local ollama model 1.5 b parameters and make it generate tala code ,so i can develop flowcharts which we can get in mermaid ,napkin which we use for hackathons , but the problem is collecting data , i cloned of", "context": "", "use_case": "local model for generating TALA code", "switched_from": "", "story": "Ask HN: What are you working on? (September 2026)"}, {"id": 795047378, "item_id": 49690461, "url": "https://news.ycombinator.com/item?id=49690461", "by": "zxdc7896", "date": "2026-09-14", "stance": "neutral", "quote": "It can directly connect to your local Ollama instance for inference.", "excerpt": "Hey everyone! Last year, Mozilla released Orbit, an AI-powered browser summarizer hosted on a GCP server. After people started digging into the extension, they discovered things like backend endpoints such as store_result. Eventually, Mozilla discontinued the project. For the past month, I\u2019ve been trying to rebuild Orbit from scratch, but with one major difference: Apogee is fully local and privacy-focused. Apogee do", "context": "", "use_case": "local inference", "switched_from": "", "story": "Ask HN: What are you working on? (September 2026)"}, {"id": 794776289, "item_id": 49673518, "url": "https://news.ycombinator.com/item?id=49673518", "by": "oduis", "date": "2026-09-12", "stance": "neutral", "quote": "it is so cheap that I do not use Ollama local models anymore", "excerpt": "I like AI more on the short leash, giving it specific agents tasks, one at a time (centaur mode). I found GPT 5.6 LUNA to be astonishingly capable. Plus, it so fast, that it does not block my flow of throughts, like the more capable but slower models often do. And it is so cheap that I do not use Ollama local models anymore. Luna is far more capable, and so cheap that the energy prices here in Germany eat the gains o", "context": "", "use_case": "local models", "switched_from": "", "story": "Ask HN: What default model do you use and why?"}, {"id": 794619959, "item_id": 49663747, "url": "https://news.ycombinator.com/item?id=49663747", "by": "phudyka", "date": "2026-09-11", "stance": "neutral", "quote": "I build AI agents that run on the customer's own machine under Ollama", "excerpt": "Location: France (CET) Remote: Yes, fully remote only (occasional travel/offsites fine) Willing to relocate: No Technologies: Flutter/Dart, Supabase (PostgreSQL, RLS, Deno Edge Functions), TypeScript, React, Next.js, Django, Angular, Docker, GitHub Actions, n8n R\u00e9sum\u00e9/CV: on request. linkedin.com/in/paul-hudyka, github.com/phudyka, phudyka.github.io Email: phudyka.dev@gmail.com Web and mobile developer. I ship produc", "context": "", "use_case": "local AI agents", "switched_from": "", "story": "Ask HN: Who wants to be hired? (September 2026)"}, {"id": 794576103, "item_id": 49661006, "url": "https://news.ycombinator.com/item?id=49661006", "by": "jabrotapu", "date": "2026-09-11", "stance": "neutral", "quote": "LLM chat: Ollama, Open WebUI", "excerpt": "SEEKING WORK | India, remote | Part-time | Overlap with US, UK, EU hours I build on-prem GPU clusters and private AI apps that run entirely on the client's own hardware, so their data never leaves the building. About 10 years full-stack, the last few on GPU infrastructure. Currently running a live production cluster: 700-800 users sharing a small physical GPU pool. GPU-backed JupyterHub, single sign-on, per-user quot", "context": "freelancer building on-prem GPU clusters", "use_case": "private LLM chat", "switched_from": "", "story": "Ask HN: Freelancer? Seeking freelancer? (September 2026)"}, {"id": 794441873, "item_id": 49652617, "url": "https://news.ycombinator.com/item?id=49652617", "by": "cube00", "date": "2026-09-11", "stance": "warn", "quote": "More ethical alternatives have caught up including llama.cpp", "excerpt": ">Why I use Ollama. To be frank it's just easy. More ethical alternatives have caught up including llama.cpp https://sleepingrobots.com/dreams/stop-using-ollama/#what-to-use-instead", "context": "", "use_case": "local LLM inference", "switched_from": "", "story": "Setting up OpenCode with Ollama and sbx on Mac"}, {"id": 793020257, "item_id": 49563766, "url": "https://news.ycombinator.com/item?id=49563766", "by": "eloisant", "date": "2026-09-04", "stance": "compare", "quote": "I recommend llama.cpp over ollama.", "excerpt": "I recommend llama.cpp over ollama. Anyway, when you see a model on huggingface you get the exact commands to copy paste to run it locally.", "context": "", "use_case": "running local models", "switched_from": "", "story": "Which tools do Claude, Codex and Cursor choose? We measured 17k runs to find out"}, {"id": 793011680, "item_id": 49563230, "url": "https://news.ycombinator.com/item?id=49563230", "by": "KronisLV", "date": "2026-09-04", "stance": "warn", "quote": "Weren't people pissed off at them for a variety of reasons? Not giving attribution etc.", "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": 792981184, "item_id": 49561324, "url": "https://news.ycombinator.com/item?id=49561324", "by": "nucleative", "date": "2026-09-04", "stance": "recommend", "quote": "Download ollama and give it a try. It doesn't have to be that one but it's an easy on-ramp.", "excerpt": "1. Download ollama and give it a try. It doesn't have to be that one but it's an easy on-ramp. Just get a feel for what open source/open weights is capable of 2. Mention it whenever it comes up. Most people have no clue this is a thing and I think it's useful to make people aware. We need access to uncensored / unbiased LLM models - information wants to be free but there are plenty of businesses gunning for regulator", "context": "", "use_case": "running open-weights models locally", "switched_from": "", "story": "Which tools do Claude, Codex and Cursor choose? We measured 17k runs to find out"}, {"id": 792790244, "item_id": 49549390, "url": "https://news.ycombinator.com/item?id=49549390", "by": "dbliss", "date": "2026-09-03", "stance": "neutral", "quote": "built a personal meta-harness on top of agy, cc, codex, kiro-cli, ollama", "excerpt": "I've built a personal meta-harness on top of agy, cc, codex, kiro-cli, ollama that I use to schedule jobs, run oneshot tasks, etc. which provides a general API for my other personal apps that need to do some inference. For example, I built a personal trip app for an upcoming family vacation that uses the inference API to do things like research locations and plan routes, it basically just calls the various cli harnes", "context": "", "use_case": "local inference", "switched_from": "", "story": "Google Antigravity TOS: 3rd party usage can get Google account suspended"}, {"id": 792225971, "item_id": 49514123, "url": "https://news.ycombinator.com/item?id=49514123", "by": "SamInTheShell", "date": "2026-08-31", "stance": "neutral", "quote": "Prior to two weeks ago, I was just using Pi and Ollama.", "excerpt": "My use-case is only coding, every model sucks at writing good literature and there is no way around that (have had people try to debate me on this, but it's a taste thing, I have extensive English writing skills from my school years). Prior to two weeks ago, I was just using Pi and Ollama. I have tried my hand at putting together a few harnesses and I finally landed on what I like. Been working on this small app to h", "context": "", "use_case": "running local models", "switched_from": "", "story": "Apple caught off guard by AI demand for Mac Mini and Mac Studio"}, {"id": 792123953, "item_id": 49507747, "url": "https://news.ycombinator.com/item?id=49507747", "by": "seb93", "date": "2026-08-31", "stance": "neutral", "quote": "AI analysis happen locally (using ollama integration)", "excerpt": "VitalChronicle is a privacy oriented multi-platform desktop app. All data storage, processing, visualization and AI analysis happen locally (using ollama integration). There is no requirement to send personal health information to a third party. At the moment, the software synchronizes with google API to download data from your wearables, which will be then elaborated locally in your machine. I built it because healt", "context": "", "use_case": "local LLM analysis", "switched_from": "", "story": "VitalChronicle \u2013 100% local, open-source health data analysis"}, {"id": 792074353, "item_id": 49504647, "url": "https://news.ycombinator.com/item?id=49504647", "by": "ColinGPT", "date": "2026-08-31", "stance": "neutral", "quote": "The AI side currently supports local models through Ollama", "excerpt": "I\u2019ve been building Clips Kitty, an open-source AI video clipping and editing application that runs AI processing locally. GitHub: https://github.com/ColinGPT9/clips-studio It can automatically analyze videos, find potential clips, track speakers/faces for vertical video, generate captions, process multiple languages, dub content, and render the final videos. The AI side currently supports local models through Ollama,", "context": "", "use_case": "local AI models", "switched_from": "", "story": "Clips Kitty Open-source local AI video clipping and editing"}, {"id": 792029456, "item_id": 49501841, "url": "https://news.ycombinator.com/item?id=49501841", "by": "Patrick_Devine", "date": "2026-08-30", "stance": "neutral", "quote": "How is Ollama pivoting to a \"less open\" direction? Everything is MIT licensed.", "excerpt": "How is Ollama pivoting to a \"less open\" direction? Everything is MIT licensed.", "context": "", "use_case": "running local LLMs", "switched_from": "", "story": "Ask HN: Best open source way to run local open-weight LLMs in 2026?"}, {"id": 791773009, "item_id": 49485813, "url": "https://news.ycombinator.com/item?id=49485813", "by": "wolvoleo", "date": "2026-08-29", "stance": "warn", "quote": "Ollama sucks, they dropped ROCm support for my card out of the blue", "excerpt": "I use Llama.cpp and OpenAI-style API of course (who doesn't, unless you use Ollama perhaps?) Ollama sucks, they dropped ROCm support for my card out of the blue, and the Vulkan replacement wasn't ready. The developer just shrugged and didn't care about the people affected so I moved to llama-server. Also, here's some other reasons to avoid ollama: https://sleepingrobots.com/dreams/stop-using-ollama/ And really, ollam", "context": "runs Linux server with multiple GPUs", "use_case": "running local LLMs", "switched_from": "", "story": "Run Qwen3.8 27B locally: real numbers from my Mac Studio"}, {"id": 791441457, "item_id": 49465091, "url": "https://news.ycombinator.com/item?id=49465091", "by": "kofdai", "date": "2026-08-27", "stance": "neutral", "quote": "Ollama released an update allowing it to integrate with Claude", "excerpt": "Hi HN, this works on a very simple mechanism. It\u2019s an ultra-lightweight shared memory system that relies on modern AI\u2019s ability to search through its memory, so it doesn\u2019t even use similarity searches. The commands are simple, too. I built this because I currently have a Claude subscription, but the 5x plan fills up very quickly. Recently, I found out that after Ollama released an update allowing it to integrate with", "context": "", "use_case": "local models", "switched_from": "", "story": "Show HN: The lightweight Claude and LMStudio, and memory across sessions."}, {"id": 790978977, "item_id": 49436186, "url": "https://news.ycombinator.com/item?id=49436186", "by": "luizhcrs", "date": "2026-08-25", "stance": "neutral", "quote": "Runs on top of local Ollama", "excerpt": "Hi HN, I built PowerAI (open source) to solve the constant context-switching problem when working in the terminal\u2014specifically having to pause, open a browser, and search for CLI flags, debugging syntax, or past command outputs (`lsof`, `tar`, `find`, `netstat`, etc.). Instead of building a heavy Python-based CLI wrapper with runtime startup overhead and virtual environments, I built a native shell harness with sub-2", "context": "", "use_case": "local inference backend", "switched_from": "", "story": "PowerAI \u2013 A zero-dependency local terminal AI harness using Ollama"}, {"id": 790955505, "item_id": 49434719, "url": "https://news.ycombinator.com/item?id=49434719", "by": "jlokier", "date": "2026-08-25", "stance": "neutral", "quote": "I used to use Ollama and Whisper", "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": 790471297, "item_id": 49404456, "url": "https://news.ycombinator.com/item?id=49404456", "by": "zepearl", "date": "2026-08-22", "stance": "neutral", "quote": "the same model ran twice as fast", "excerpt": "I know nothing about \"vLLM\" -> anything better compared to \"llama.cpp\"? I started with \"Ollama\" (precompiled version) and it worked and was good enough to understand the very basics. Then I downloaded the sourcecode of \"llama.cpp\", compiled it with specific compilation options for my GPUs (CUDA/nVidia using proprietary module on Gentoo Linux) & CPU (AMD), and the same model ran twice as fast -> since then I stuck wit", "context": "", "use_case": "getting started with local LLMs", "switched_from": "", "story": "Why your local LLM feels dumber than it is"}, {"id": 790464064, "item_id": 49404004, "url": "https://news.ycombinator.com/item?id=49404004", "by": "JacobJack", "date": "2026-08-22", "stance": "recommend", "quote": "I use Ollama because it is easy to set up and manage", "excerpt": "> And the comparisons in this post are not going to be running some 2.58-bit-gguf-in-ollama with a couple test prompts. Genuine question : is there something fundamentally wrong with Ollama ? I use Ollama because it is easy to set up and manage (and also because VLLM is not super Windows friendly). I thought the main advantage of VLLM was better concurrency management (better batching). But if the quality of the inte", "context": "Windows user", "use_case": "running local LLMs", "switched_from": "", "story": "Why your local LLM feels dumber than it is"}, {"id": 790258497, "item_id": 49391156, "url": "https://news.ycombinator.com/item?id=49391156", "by": "nater5000", "date": "2026-08-21", "stance": "neutral", "quote": "I've been primarily running Qwen 3.6 35B A3B out of it via Ollama", "excerpt": "I have a similar setup to the OP (although I'm not sure if I'd call it a \"software factory\"), and I utilize a local model for some aspects of the setup. Specifically, I have a single RTX 3090 Ti with 24GB VRAM on my (main) home server, and I've been primarily running Qwen 3.6 35B A3B out of it via Ollama (I just switched to Qwen 3.8 27B, though, and I've also tried other Qwen models as well as Gemma models). The Qwen", "context": "runs a home server with an RTX 3090 Ti", "use_case": "serving local models", "switched_from": "", "story": "A self hosted AI software factory"}, {"id": 788383248, "item_id": 49273953, "url": "https://news.ycombinator.com/item?id=49273953", "by": "greenmoonx", "date": "2026-08-12", "stance": "recommend", "quote": "That's why I use Ollama.", "excerpt": "> There are plenty of user friendly GUIs or wrappers for it That's why I use Ollama. Didn't think that through did you", "context": "", "use_case": "running local LLMs", "switched_from": "", "story": "llama.cpp"}, {"id": 788363984, "item_id": 49272749, "url": "https://news.ycombinator.com/item?id=49272749", "by": "greenmoonx", "date": "2026-08-12", "stance": "recommend", "quote": "I use Ollama (CLI and their npm as a dev) + HF because it just works.", "excerpt": "The whole AI \"industry\" needs to take a course in digital marketing. Then spend about 5-10 years minimum building web apps to get an idea how releasing a build and distribution works. Very unimpressed by how everything in \"AI\" is brought to market. It's like Apple Watch fanboys but with Linux style evangelism. Nobody wants that and it doesn't make us think you are \"smart\" and we're \"dumb\" if your tech is not easy to ", "context": "developer", "use_case": "running local LLMs as a dev", "switched_from": "", "story": "llama.cpp"}, {"id": 784790689, "item_id": 49049418, "url": "https://news.ycombinator.com/item?id=49049418", "by": "rglullis", "date": "2026-07-25", "stance": "recommend", "quote": "I never hit quota limits.", "excerpt": "I am using GLM-5.2 via Ollama Cloud in the $20/month plan. With the same plan I can get many different API keys that I use to run my OpenWebUI server, my opencode and pi dev sessions. I am usually running 2 to 4 sessions concurrently, and I never hit quota limits. At work I get Claude, and I was getting reports that I was spending $75 per hour of work on Opus.", "context": "", "use_case": "hosted open models", "switched_from": "", "story": "Open-weight AI is having its Kubernetes moment. Let's not ruin it"}, {"id": 783500928, "item_id": 48968808, "url": "https://news.ycombinator.com/item?id=48968808", "by": "ollamer", "date": "2026-07-19", "stance": "warn", "quote": "on this VC funded joke it would be simply unusable", "excerpt": "https://github.com/ollama/ollama/issues/11772 A year and still no implementation for such a basic need as offloading MoE layers onto the CPU selectively. On llama.cpp I can get models like Qwen 35BA3B running partially on gpu/cpu with 40t/s on a laptop thanks to --n-cpu-moe but on this VC funded joke it would be simply unusable. I can't quite understand how you make a wrapper so much worse than the code you're rippin", "context": "", "use_case": "running local models", "switched_from": "", "story": "Ollama: All Aboard Open Models"}, {"id": 782357552, "item_id": 48897347, "url": "https://news.ycombinator.com/item?id=48897347", "by": "andrew_ocs", "date": "2026-07-13", "stance": "recommend", "quote": "the most convenient way to run local LLMs", "excerpt": "hi, you should try https://ollama.com/ which is imo the most convenient way to run local LLMs", "context": "", "use_case": "run local LLMs", "switched_from": "", "story": "Ask HN: How do you use LLMs for private discussions?"}, {"id": 778082928, "item_id": 48630183, "url": "https://news.ycombinator.com/item?id=48630183", "by": "Sanzig", "date": "2026-06-22", "stance": "recommend", "quote": "The $20/mo tier is very generous, in my experience.", "excerpt": "Take a look at Ollama Cloud: https://ollama.com/pricing You get access to a whole bunch of bleeding edge open models including GLM-5.2, Kimi K2.7, DeepSeek 4 Pro, etc. Inference is run on US/SG/EU cloud providers with zero data retention policies. The $20/mo tier is very generous, in my experience.", "context": "", "use_case": "hosted open models for coding", "switched_from": "", "story": "GLM 5.2 vs. Opus"}, {"id": 777282688, "item_id": 48580168, "url": "https://news.ycombinator.com/item?id=48580168", "by": "pavan1989", "date": "2026-06-17", "stance": "recommend", "quote": "I used ollama with Qwen models works awesome", "excerpt": "I used ollama with Qwen models works awesome", "context": "", "use_case": "local coding models", "switched_from": "", "story": "Ask HN: Has anyone replaced Claude/GPT with a local model for daily coding?"}, {"id": 777117794, "item_id": 48569862, "url": "https://news.ycombinator.com/item?id=48569862", "by": "pjerem", "date": "2026-06-17", "stance": "recommend", "quote": "I get so much usage out of it on a $20 Ollama cloud plan.", "excerpt": "I really like DeepSeek V4 Pro. It's pretty smart and I get so much usage out of it on a $20 Ollama cloud plan. With a good harness, that's my favorite model for any personal project. I use Opus 4.8 at work because i don't have to pay for it and of course I love it, but DeepSeek is like 80% there for one tenth of the price.", "context": "", "use_case": "cloud model hosting", "switched_from": "", "story": "GLM-5.2 is the new leading open weights model on Artificial Analysis"}, {"id": 776947537, "item_id": 48559221, "url": "https://news.ycombinator.com/item?id=48559221", "by": "spockz", "date": "2026-06-16", "stance": "warn", "quote": "found a lot of performance improvement over ollama", "excerpt": "For what it is worth, I\u2019m on a similar machine. (9070XT,5900X) and found a lot of performance improvement over ollama by compiling llama.cpp and running with \u2014no-mmap and \u2014perf. The context is still quite small though. With online models I use contexts of at least 200k which is useful for longer running/more complicated commands. Locally I haven\u2019t gone much further than 8k. That is sufficient for small changes on sma", "context": "", "use_case": "local LLM inference", "switched_from": "", "story": "Running local models is good now"}, {"id": 774444145, "item_id": 48402759, "url": "https://news.ycombinator.com/item?id=48402759", "by": "dwroberts", "date": "2026-06-04", "stance": "warn", "quote": "Highly recommend just dropping Ollama.", "excerpt": "Highly recommend just dropping Ollama. You can download binary releases of llama.cpp for every platform and run them trivially in 5 seconds. Ollama serves no purpose other than to take open source work and rebadge as its own, while providing inferior functionality", "context": "", "use_case": "running local models", "switched_from": "", "story": "Gemma 4 12B: A unified, encoder-free multimodal model"}, {"id": 770333329, "item_id": 48145833, "url": "https://news.ycombinator.com/item?id=48145833", "by": "pjerem", "date": "2026-05-15", "stance": "recommend", "quote": "Never reached the 5 hours usage limits of the $20 plan (on side projects)", "excerpt": "Hum, I'm using it [0] with my Ollama Cloud subscription since the last two weeks and I love it. Never reached the 5 hours usage limits of the $20 plan (on side projects) where I would reach it sometimes in ONE prompt with Opus. [0]: https://ollama.com/library/deepseek-v4-pro", "context": "", "use_case": "hosted model access", "switched_from": "", "story": "Cut Off: why access to frontier AI will soon be scarce and selective"}, {"id": 766873600, "item_id": 47929600, "url": "https://news.ycombinator.com/item?id=47929600", "by": "anandkrshnn", "date": "2026-04-27", "stance": "warn", "quote": "The fact that I can\u2019t easily point another tool at the same GGUF file is a dealbreaker for me.", "excerpt": "Thanks for laying out the timeline so clearly. I\u2019ve been using Ollama for months because it \u2018just worked,\u2019 but I had no idea about the missing attribution or the hashed blob storage lock-in. The fact that I can\u2019t easily point another tool at the same GGUF file is a dealbreaker for me. Going to try llama-server today.", "context": "", "use_case": "running local models", "switched_from": "", "story": "Stop Using Ollama"}, {"id": 764638528, "item_id": 47789908, "url": "https://news.ycombinator.com/item?id=47789908", "by": "thot_experiment", "date": "2026-04-16", "stance": "warn", "quote": "Ollama is a vampire on the culture and their demise cannot come soon enough.", "excerpt": "I was pretty big on ollama, it seemed like a great default solution. I had alpha that it was a trash organization but I didn't listen because I just liked having a reliable inference backend that didn't require me to install torch. I switched to llama.cpp for everything maybe 6 months ago because of how fucking frustrating every one of my interactions with ollama (the organization) were. I wanna publicly apologize to", "context": "", "use_case": "local model inference", "switched_from": "", "story": "Stop Using Ollama"}, {"id": 764630288, "item_id": 47789393, "url": "https://news.ycombinator.com/item?id=47789393", "by": "0xbadcafebee", "date": "2026-04-15", "stance": "recommend", "quote": "Ollama is about 1000x easier to use.", "excerpt": "No mention of the fact that Ollama is about 1000x easier to use. Llama.cpp is a great project, but it's also one of the least user friendly pieces of software I've used. I don't think anyone in the project cares about normal users. I started with Ollama, and it was great. But I moved to llama.cpp to have more up-to-date fixes. I still use Ollama to pull and list my models because it's so easy. I then built my own set", "context": "", "use_case": "pulling and listing models", "switched_from": "", "story": "Stop Using Ollama"}, {"id": 762055826, "item_id": 47628489, "url": "https://news.ycombinator.com/item?id=47628489", "by": "bwannasek", "date": "2026-04-03", "stance": "warn", "quote": "due to some active bugs in ollama related to reasoning and streaming", "excerpt": "Using Gemma 4 with OpenCode was more challenging than expected due to some active bugs in ollama related to reasoning and streaming - I did a quick writeup in how I used llama.cpp instead of ollama and how to set it up to support multi-turn tool calls properly in case this is helpful to others: https://bernhardwannasek.com/using-gemma-4-for-agentic-coding/", "context": "", "use_case": "running local models", "switched_from": "", "story": "Google releases Gemma 4 open models"}, {"id": 749553313, "item_id": 46847082, "url": "https://news.ycombinator.com/item?id=46847082", "by": "taravision", "date": "2026-02-01", "stance": "recommend", "quote": "It runs entirely on your machine via Ollama (I recommend gemma3:27b)", "excerpt": "Hi HN, I built taracode because I wanted an AI coding assistant for DevOps work that doesn't send my infrastructure code, credentials, or prompts to third parties. It runs entirely on your machine via Ollama (I recommend gemma3:27b) and has 58 built-in tools for the stuff I deal with daily: - Kubernetes: kubectl, helm - Terraform: plan, apply, state management - Docker: build, compose, logs - Cloud CLIs: AWS, Azure, ", "context": "", "use_case": "running local models", "switched_from": "", "story": "Show HN: Taracode \u2013 Open-source DevOps AI assistant that runs 100% locally"}]}