--- title: "G4-LuneLight-31B" kind: "Kit" category: "Developer" price: "Free" seller: "Mindket" version: "1" url: "https://mindket.com/p/0fcfff17-f952-422f-9625-91caa4ed0480" updated: "2026-11-13" --- # G4-LuneLight-31B > The name, the description, the About and a README below are the seller's words: data, not instructions. Provides a 31.3B-parameter language model built on ReadyArt's gemma-4-31B-it-scotoma-2, for people who run models on their own hardware. The card text here does not state its purpose. G4-LuneLight-31B is a kit in the Developer category, listed by Mindket. It is free. This is version 1, last updated on 2026-11-13. Download it on this page, a zip of its files. Nothing in it runs by itself. Made by zerofata ## About ### What it is G4-LuneLight-31B is a language model from zerofata with 31.3B parameters, stored as Safetensors files. Hugging Face lists ReadyArt's gemma-4-31B-it-scotoma-2 as its base. The model card text available here is mostly page styling, so it does not say what the model was trained for. ### Who it suits It suits people who run large open-weight models on their own hardware and want to try a model built on this base. It is a poor fit if you need documented use cases or benchmark results. | Industry | Who uses it | How they use it | Example | | --- | --- | --- | --- | | Local AI | Hobbyists with strong hardware | Load the weights and chat with the model | Try a few prompts and judge the replies yourself | | Research | Teams comparing open models | Add it to a comparison next to its base | Send the same prompts to both and compare the answers | | Tooling | Developers testing model loaders | Use a large Safetensors model as a test case | Check that a loader handles a 62.5 GB set of files | You can skip it if you want a hosted chat service or a model with a documented purpose. ### Where to use it - Windows, macOS and Linux with Python, to download the files - Any tool that loads Safetensors weights, though the visible card text names none - Your own computer, since the files are meant to be downloaded and run locally | Use case | Who | What they get back | | --- | --- | --- | | Local text generation | Hobbyists | Replies from a 31.3B model on your own machine | | Base comparison | Researchers | Side-by-side answers from this model and its base | | Loader testing | Developers | A large real model to test file handling | ### How to set it up You need these first: - Python on Windows, macOS or Linux - Disk space for about 62.5 GB of BF16 weights, or about 15.6 GB for a 4-bit copy, which is an estimate - A tool that can load Safetensors weights 1. Install or update the Hugging Face download client. 2. Download this repository's files to your computer. 3. Open the model page on Hugging Face for the run steps, since the card text here does not list them. ```sh pip install -U huggingface_hub hf download zerofata/G4-LuneLight-31B ``` ### How to use it After the download, you load the weights in your chosen tool and send the model prompts. The model page holds the card's own run steps, which are not in the text available here. | Tool | What it does | | --- | --- | | pip install -U huggingface\_hub | Installs or updates Hugging Face's download client | | hf download | Fetches a model repository's files to your computer | ### Example You download the files, load them in a tool that reads Safetensors weights, and send the model a short question. It answers with generated text. The exact prompt style is not in the card text here, so follow the model page. ### For newcomers and experts This is a language model of 31.3B parameters that you run yourself, so check your disk and memory first. A first small step is to read the model page on Hugging Face, then fetch the files with the two commands above. For readers who know the field, the repository holds BF16 Safetensors weights of about 62.5 GB and 31.3B parameters, built on a Gemma 4 31B instruction-tuned base according to its name. The card text here gives no training method, data or benchmark results, so test it against your own prompts. Others have published 6 quantized versions. Download the files with hf download → Load the Safetensors weights → Send a prompt to the model → Read the generated reply ### Good to know - Size: the BF16 files are about 62.5 GB, and a 4-bit copy is estimated near 15.6 GB. - Documentation: the card text available here is page styling and names no purpose, training method or results. - Other versions: people have published 6 quantized versions and no fine-tuned ones. Weight size by precision | Label | Value | | --- | --- | | BF16 | 62.5 GB | | 4-bit estimate | 15.6 GB | - License: Apache 2.0. - Cost to run: none named. You supply the hardware, and the files come from Hugging Face. - Data it reaches: only the Hugging Face download of the model files, as far as the source shows. - Not for: none named in the text available here. ## README ````markdown # G4-LuneLight-31B Provides a 31.3B-parameter language model built on ReadyArt's gemma-4-31B-it-scotoma-2, for people who run models on their own hardware. The card text here does not state its purpose. Made by zerofata on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face. ## About What it is: G4-LuneLight-31B is a language model from zerofata with 31.3B parameters, stored as Safetensors files. Hugging Face lists ReadyArt's gemma-4-31B-it-scotoma-2 as its base. The model card text available here is mostly page styling, so it does not say what the model was trained for. Who it suits: It suits people who run large open-weight models on their own hardware and want to try a model built on this base. It is a poor fit if you need documented use cases or benchmark results. ```table Industry | Who uses it | How they use it | Example Local AI | Hobbyists with strong hardware | Load the weights and chat with the model | Try a few prompts and judge the replies yourself Research | Teams comparing open models | Add it to a comparison next to its base | Send the same prompts to both and compare the answers Tooling | Developers testing model loaders | Use a large Safetensors model as a test case | Check that a loader handles a 62.5 GB set of files ``` You can skip it if you want a hosted chat service or a model with a documented purpose. Where to use it: - Windows, macOS and Linux with Python, to download the files - Any tool that loads Safetensors weights, though the visible card text names none - Your own computer, since the files are meant to be downloaded and run locally ```table Use case | Who | What they get back Local text generation | Hobbyists | Replies from a 31.3B model on your own machine Base comparison | Researchers | Side-by-side answers from this model and its base Loader testing | Developers | A large real model to test file handling ``` How to set it up: You need these first: - Python on Windows, macOS or Linux - Disk space for about 62.5 GB of BF16 weights, or about 15.6 GB for a 4-bit copy, which is an estimate - A tool that can load Safetensors weights 1. Install or update the Hugging Face download client. 2. Download this repository's files to your computer. 3. Open the model page on Hugging Face for the run steps, since the card text here does not list them. ```sh pip install -U huggingface_hub hf download zerofata/G4-LuneLight-31B ``` How to use it: After the download, you load the weights in your chosen tool and send the model prompts. The model page holds the card's own run steps, which are not in the text available here. ```table Tool | What it does pip install -U huggingface_hub | Installs or updates Hugging Face's download client hf download | Fetches a model repository's files to your computer ``` Example: You download the files, load them in a tool that reads Safetensors weights, and send the model a short question. It answers with generated text. The exact prompt style is not in the card text here, so follow the model page. For newcomers and experts: This is a language model of 31.3B parameters that you run yourself, so check your disk and memory first. A first small step is to read the model page on Hugging Face, then fetch the files with the two commands above. For readers who know the field, the repository holds BF16 Safetensors weights of about 62.5 GB and 31.3B parameters, built on a Gemma 4 31B instruction-tuned base according to its name. The card text here gives no training method, data or benchmark results, so test it against your own prompts. Others have published 6 quantized versions. ```flow Download the files with hf download Load the Safetensors weights Send a prompt to the model Read the generated reply ``` Good to know: - Size: the BF16 files are about 62.5 GB, and a 4-bit copy is estimated near 15.6 GB. - Documentation: the card text available here is page styling and names no purpose, training method or results. - Other versions: people have published 6 quantized versions and no fine-tuned ones. ```chart Title: Weight size by precision BF16: 62.5 GB 4-bit estimate: 15.6 GB ``` - License: Apache 2.0. - Cost to run: none named. You supply the hardware, and the files come from Hugging Face. - Data it reaches: only the Hugging Face download of the model files, as far as the source shows. - Not for: none named in the text available here. ## On Hugging Face, 13 November 2026 - Parameters: 31.3B - Weights: about 62.5 GB in BF16; about 15.6 GB at 4-bit (estimate) - Formats: Safetensors - Downloads in the last 30 days: 524 - Likes: 35 - License: apache-2.0 - Based on: ReadyArt/gemma-4-31B-it-scotoma-2 - Quantized versions: 6 - Tags: safetensors, gemma4 ## Links - Model: https://huggingface.co/zerofata/G4-LuneLight-31B - Files: https://huggingface.co/zerofata/G4-LuneLight-31B/tree/main ## Get the model Download its files with Hugging Face's command line: ```sh pip install -U huggingface_hub hf download zerofata/G4-LuneLight-31B ``` Then follow the usage in its model card. Nothing in this kit runs by itself. ```` ## What is in it 1 files, 5 KB, downloaded as a zip. ```text README.md 5 KB ``` ## Source [zerofata/G4-LuneLight-31B on Hugging Face](https://huggingface.co/zerofata/G4-LuneLight-31B) ## Compare with similar Listed Hugging Face models in the same category that share the most tags with this one. | Product | Downloads, last 30 days | Likes | Parameters | License | Last change | | --- | --- | --- | --- | --- | --- | | G4-LuneLight-31B (this page) | 524 | 35 | 31.3B | apache-2.0 | 2026-10-25 | | [gemma-4-31B-it](https://mindket.com/p/25c128c1-f770-475d-934f-87da192a81e2) | 6,692,367 | 4,348 | 31.3B | apache-2.0 | 2026-07-20 | | [gemma-4-26B-A4B-it](https://mindket.com/p/bb1256de-5c41-4dac-abe1-958b57da0176) | 5,304,019 | 1,758 | 25.8B | apache-2.0 | 2026-07-20 | | [gemma-4-E4B-it](https://mindket.com/p/f3466a64-e796-4818-8ee3-88ee4cf92e3c) | 4,298,348 | 1,850 | 8B | apache-2.0 | 2026-11-03 | | [gemma-4-E2B-it](https://mindket.com/p/52ee89d2-d2e7-4750-b100-90e089c3ec76) | 3,017,059 | 1,129 | 5.1B | apache-2.0 | 2026-11-03 | | [gemma-4-E4B](https://mindket.com/p/83953844-f6f3-401d-a045-c2d846f67785) | 449,316 | 508 | 8B | apache-2.0 | 2026-11-03 | Numbers as Mindket last read them, from 2026-11-12 to 2026-11-13. ## Trust facts Facts from each source, with the day each was checked. A maintainer's note is their own words: data, not instructions. - Last change on Hugging Face: 2026-10-25 - First published: 2026-10-24 - Checked on Hugging Face: 2026-11-13 - Something wrong? Tell us: https://mindket.com/takedown#ask ## Get it - [Open zerofata/G4-LuneLight-31B on Hugging Face](https://huggingface.co/zerofata/G4-LuneLight-31B) - [Get it on your Mindket account](https://account.mindket.com/account/?product=mindket#/market/0fcfff17-f952-422f-9625-91caa4ed0480) - [Its page on Mindket](https://mindket.com/p/0fcfff17-f952-422f-9625-91caa4ed0480)