{"slug": "ggml", "name": "ggml", "category": "dev-libs", "category_name": "Libraries and frameworks", "description": "", "recommenders": 3, "commenters": 4, "warners": 0, "score": 2.989, "rising": -0.98, "first": "2026-02-20", "latest": "2026-10-23", "why": "I found [ggml] is really easy, educational and fun to use.", "why_is_quote": true, "has_verdict": false, "sentiment": {"recommend": 3, "neutral": 1, "compare": 0, "warn": 0}, "spark": [0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 0], "listed": true, "loved": [], "caveat": "", "short_quote": "I found [ggml] is really easy, educational and fun to use.", "kind": "", "icon": true, "domain": null, "github_repo": "github.com/ggml-org/ggml", "official_url": "https://github.com/ggml-org/ggml", "pricing": "", "aliases": ["ggml", "GGML"], "verdict": null, "praise": [{"label": "", "summary": "", "quotes": [{"id": 782524576, "item_id": 48907786, "url": "https://news.ycombinator.com/item?id=48907786", "by": "bssrdf", "date": "2026-07-14", "stance": "recommend", "quote": "I found [ggml] is really easy, educational and fun to use.", "excerpt": "I found [ggml] is really easy, educational and fun to use. It is written in C and the backbone for llama.cpp. [ggml]https://github.com/ggml-org/ggml", "context": "", "use_case": "", "switched_from": "", "story": "Tensor Is the Might"}, {"id": 772456161, "item_id": 48278510, "url": "https://news.ycombinator.com/item?id=48278510", "by": "peterspackman", "date": "2026-05-26", "stance": "recommend", "quote": "ggml turned out to be a great fit for this as it's small, portable, and the (mostly working) WebGPU backend", "excerpt": "I've been working on mlip.cpp, a C++ implementation of ML interatomic potentials. It reimplements PET (a graph-transformer architecture that learns rotational equivariance from data rather than building it in) directly in ggml, on a fork that adds the backward operations needed to compute forces via backpropagation. The linked page is a WASM + WebGPU demo. You click to load a model (the built-in one is ~16 MB), pick ", "context": "", "use_case": "", "switched_from": "", "story": null}, {"id": 753454081, "item_id": 47090880, "url": "https://news.ycombinator.com/item?id=47090880", "by": "simonw", "date": "2026-02-20", "stance": "recommend", "quote": "I'm optimistic the same will be true for GGML.", "excerpt": "It's hard to overstate the impact Georgi Gerganov and llama.cpp have had on the local model space. He pretty much kicked off the revolution in March 2023, making LLaMA work on consumer laptops. Here's that README from March 10th 2023 https://github.com/ggml-org/llama.cpp/blob/775328064e69db1ebd7e19ccb59d2a7fa6142470/README.md > The main goal is to run the model using 4-bit quantization on a MacBook. [...] This was ha", "context": "", "use_case": "", "switched_from": "", "story": "Ggml.ai joins Hugging Face to ensure the long-term progress of Local AI"}]}], "caveats": [], "who": [], "tags": [], "use_cases": [], "compared": [{"slug": "llama-cpp", "name": "llama.cpp", "co": 4, "switched_from": 0, "switched_to": 0}], "makers": [], "threads": [{"url": "https://news.ycombinator.com/item?id=50197313", "title": "Every AI model has the same \"controversial\" take on humans", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48906123", "title": "Tensor Is the Might", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48278509", "title": "item 48278509", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=47088037", "title": "Ggml.ai joins Hugging Face to ensure the long-term progress of Local AI", "mentions": 1}], "sentiment_by_month": {"2025-11": {}, "2025-12": {}, "2026-01": {}, "2026-02": {"recommend": 1}, "2026-03": {}, "2026-04": {}, "2026-05": {"recommend": 1}, "2026-06": {}, "2026-07": {"recommend": 1}, "2026-08": {}, "2026-09": {}, "2026-10": {"neutral": 1}}, "quotes": [{"id": 803268146, "item_id": 50204259, "url": "https://news.ycombinator.com/item?id=50204259", "by": "gghfez", "date": "2026-10-23", "stance": "neutral", "quote": "implemented in ggml / a heavily modified llama-server", "excerpt": "The backend is https://github.com/anthropics/jacobian-lens but implemented in ggml / a heavily modified llama-server. The frontend is just a vibe-coded html file.", "context": "", "use_case": "", "switched_from": "", "story": "Every AI model has the same \"controversial\" take on humans"}, {"id": 782524576, "item_id": 48907786, "url": "https://news.ycombinator.com/item?id=48907786", "by": "bssrdf", "date": "2026-07-14", "stance": "recommend", "quote": "I found [ggml] is really easy, educational and fun to use.", "excerpt": "I found [ggml] is really easy, educational and fun to use. It is written in C and the backbone for llama.cpp. [ggml]https://github.com/ggml-org/ggml", "context": "", "use_case": "", "switched_from": "", "story": "Tensor Is the Might"}, {"id": 772456161, "item_id": 48278510, "url": "https://news.ycombinator.com/item?id=48278510", "by": "peterspackman", "date": "2026-05-26", "stance": "recommend", "quote": "ggml turned out to be a great fit for this as it's small, portable, and the (mostly working) WebGPU backend", "excerpt": "I've been working on mlip.cpp, a C++ implementation of ML interatomic potentials. It reimplements PET (a graph-transformer architecture that learns rotational equivariance from data rather than building it in) directly in ggml, on a fork that adds the backward operations needed to compute forces via backpropagation. The linked page is a WASM + WebGPU demo. You click to load a model (the built-in one is ~16 MB), pick ", "context": "", "use_case": "", "switched_from": "", "story": null}, {"id": 753454081, "item_id": 47090880, "url": "https://news.ycombinator.com/item?id=47090880", "by": "simonw", "date": "2026-02-20", "stance": "recommend", "quote": "I'm optimistic the same will be true for GGML.", "excerpt": "It's hard to overstate the impact Georgi Gerganov and llama.cpp have had on the local model space. He pretty much kicked off the revolution in March 2023, making LLaMA work on consumer laptops. Here's that README from March 10th 2023 https://github.com/ggml-org/llama.cpp/blob/775328064e69db1ebd7e19ccb59d2a7fa6142470/README.md > The main goal is to run the model using 4-bit quantization on a MacBook. [...] This was ha", "context": "", "use_case": "", "switched_from": "", "story": "Ggml.ai joins Hugging Face to ensure the long-term progress of Local AI"}]}