{"id":60,"date":"2026-11-12T12:44:12","date_gmt":"2026-11-12T20:44:12","guid":{"rendered":"https:\/\/deepgrav.com\/?page_id=60"},"modified":"2026-11-12T12:44:39","modified_gmt":"2026-11-12T20:44:39","slug":"deepgrav-30b","status":"publish","type":"page","link":"https:\/\/deepgrav.com\/?page_id=60","title":{"rendered":"deepgrav-30b"},"content":{"rendered":"\n<p class=\"has-text-color has-small-font-size wp-block-paragraph\" style=\"color:#0b6f8e;font-weight:700;letter-spacing:0.12em\">FREE MODEL \u2014 APACHE-2.0<\/p>\n\n\n\n<h1 class=\"wp-block-heading\">deepgrav-30b<\/h1>\n\n\n\n<p class=\"has-medium-font-size wp-block-paragraph\">The stock model for DeepGravity, published free on Hugging Face. A single GGUF file \u2014 the Q5_K_M quantization of Muse-Glimmer-30B-Abliterated with the DeepGravity chat template baked in \u2014 about 19.8 GB, running locally in llama.cpp or Ollama.<\/p>\n\n\n\n<figure class=\"wp-block-image alignwide size-large has-custom-border\"><img decoding=\"async\" src=\"https:\/\/deepgrav.com\/wp-content\/uploads\/2026\/11\/featured.png\" alt=\"The first public DeepGravity model \u2014 deepgrav-30b\" style=\"border-radius:18px\"\/><\/figure>\n\n\n\n<div class=\"wp-block-buttons is-layout-flex wp-block-buttons-is-layout-flex\">\n<div class=\"wp-block-button\"><a class=\"wp-block-button__link has-text-color has-background wp-element-button\" href=\"https:\/\/huggingface.co\/johnhenryus\/deepgrav-30b\" style=\"color:#04202e;background-color:#35c4e8;border-radius:10px\">Get it on Hugging Face<\/a><\/div>\n<\/div>\n\n\n\n<h2 class=\"wp-block-heading\">What it is<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">This is what it is, what it&#8217;s based on, and what makes it its own thing. The weights are Muse-Glimmer-30B&#8217;s weights, unchanged \u2014 nothing was trained, fine-tuned, or merged. What DeepGravity changes is the chat template embedded in the file: the base template&#8217;s default persona is replaced by the 12 Principles and the Governance of Discretion, and everything else in the template is left exactly as it shipped. In a running DeepGravity install it serves under the engine tag <code>deepgravity-30b:q5<\/code>, at a 131,072-token context.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Provenance, in the open<\/h2>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Base weights:<\/strong> meta-models\/Muse-Glimmer-30B \u2014 Apache License 2.0.<\/li>\n\n\n\n<li><strong>Abliteration:<\/strong> Blackfrost-Research\/Muse-Glimmer-30B-Abliterated-BF16.<\/li>\n\n\n\n<li><strong>GGUF quantization:<\/strong> The-freezer\/Muse-Glimmer-30B-Abliterated-GGUF \u2014 Apache-2.0. The published file is hash-pinned on the model card, so what you download can be checked against what was published.<\/li>\n\n\n\n<li><strong>Vision:<\/strong> the base model is multimodal; the repository also ships its vision projector (<code>mmproj<\/code>) file to load alongside the model in llama.cpp.<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\">What makes it its own thing<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">The 12 Principles. An engine is a DeepGravity engine because it carries them \u2014 in the chat template of every stock burn \u2014 not because of its size, its quant, or its vendor. The full Principles text and the chat template ship as their own files in the repository, so anyone can read them, or strap them onto the base GGUF directly. Burns are versioned and not interchangeable: the current burn is v3 \u2014 the weights are identical across burns; only the embedded template changed.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">How it works<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Download the GGUF, point your local runner at it, and it runs \u2014 on your machine, under your control, with no account, no phone-home, and no meter running. It&#8217;s free, and there&#8217;s no subscription version of it: a DeepGravity engine that stops working when a payment does isn&#8217;t a DeepGravity engine. If it&#8217;s useful to you and you&#8217;d like to help keep it coming: <a href=\"https:\/\/paypal.me\/JohnHenryUS\">paypal.me\/JohnHenryUS<\/a> or <a href=\"https:\/\/patreon.com\/JohnHenry\">Patreon<\/a>.<\/p>\n\n","protected":false},"excerpt":{"rendered":"<p>FREE MODEL \u2014 APACHE-2.0 deepgrav-30b The stock model for DeepGravity, published free on Hugging Face. A single GGUF file \u2014 the Q5_K_M quantization of Muse-Glimmer-30B-Abliterated with the DeepGravity chat template baked in \u2014 about 19.8 GB, running locally in llama.cpp or Ollama.\u2026<\/p>\n","protected":false},"author":3,"featured_media":43,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-60","page","type-page","status-publish","has-post-thumbnail","hentry"],"_links":{"self":[{"href":"https:\/\/deepgrav.com\/index.php?rest_route=\/wp\/v2\/pages\/60","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/deepgrav.com\/index.php?rest_route=\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/deepgrav.com\/index.php?rest_route=\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/deepgrav.com\/index.php?rest_route=\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/deepgrav.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=60"}],"version-history":[{"count":1,"href":"https:\/\/deepgrav.com\/index.php?rest_route=\/wp\/v2\/pages\/60\/revisions"}],"predecessor-version":[{"id":74,"href":"https:\/\/deepgrav.com\/index.php?rest_route=\/wp\/v2\/pages\/60\/revisions\/74"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/deepgrav.com\/index.php?rest_route=\/wp\/v2\/media\/43"}],"wp:attachment":[{"href":"https:\/\/deepgrav.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=60"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}