[{"data":1,"prerenderedAt":130},["ShallowReactive",2],{"learn-llm-model-licenses":3},{"collection":4,"slug":5,"path":6,"title":7,"seoTitle":8,"description":9,"summary":10,"primaryKeyword":11,"cluster":12,"published":13,"updated":13,"reviewed":13,"readingMinutes":14,"html":15,"toc":16,"wordCount":55,"sources":56,"faq":92,"related":105,"tested":122,"internalLinks":123},"learn","llm-model-licenses","\u002Flearn\u002Fllm-model-licenses\u002F","Open-weight model licenses compared","Open-Weight Model License Comparison: MIT, Apache, Llama","Commercial use, notices, naming rules, user thresholds and use policies under MIT, Apache 2.0, Llama 3.3, Llama 3.1 and Gemma terms, and what variants inherit.","GLM models use MIT; Gemma 4, Qwen3.8 and gpt-oss-20b use Apache 2.0. Both let you use, modify and sell the weights if you meet their notice conditions. The Llama 3.3 and 3.1 licenses add attribution, a naming rule, a 700-million-user threshold and an acceptable use policy. Gemma 3's terms add a prohibited use policy that must travel with every copy. A refusal-reduced variant carries its base model's terms. Not legal advice.","open weight model license comparison","licensing","2026-11-14",7,"\u003Cblockquote>\n\u003Cp>\u003Cstrong>Not legal advice.\u003C\u002Fstrong> This page summarizes license texts as published on 14 November 2026, with short quotes. Licenses and use policies change, and how a clause applies depends on your facts and jurisdiction. For commercial or legal decisions, have a lawyer read the full terms.\u003C\u002Fp>\n\u003C\u002Fblockquote>\n\u003Ch2 id=\"which-license-covers-which-model\">Which license covers which model\u003C\u002Fh2>\n\u003Cp>The base models covered on this site fall under five licenses. The license fields below were checked on each base model’s Hugging Face repository on 14 November 2026.\u003C\u002Fp>\n\u003Cdiv class=\"table-wrap\" role=\"region\" aria-label=\"Scrollable table\" tabindex=\"0\">\n\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth>Base model\u003C\u002Fth>\n\u003Cth>License\u003C\u002Fth>\n\u003Cth>Variants on this site\u003C\u002Fth>\n\u003C\u002Ftr>\n\u003C\u002Fthead>\n\u003Ctbody>\n\u003Ctr>\n\u003Ctd>GLM-5.3-Flash\u003C\u002Ftd>\n\u003Ctd>MIT\u003C\u002Ftd>\n\u003Ctd>\u003Ca href=\"\u002Fmodels\u002Fglm-5-3-flash-uncensored\u002F\">GLM-5.3-Flash variants\u003C\u002Fa>\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>GLM-4.7-Flash\u003C\u002Ftd>\n\u003Ctd>MIT\u003C\u002Ftd>\n\u003Ctd>\u003Ca href=\"\u002Fmodels\u002Fglm-4-7-flash-uncensored\u002F\">GLM-4.7-Flash variants\u003C\u002Fa>\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Gemma 4 26B A4B\u003C\u002Ftd>\n\u003Ctd>Apache 2.0\u003C\u002Ftd>\n\u003Ctd>\u003Ca href=\"\u002Fmodels\u002Fgemma-4-26b-a4b-uncensored\u002F\">Gemma 4 variants\u003C\u002Fa>\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Qwen3.8 27B\u003C\u002Ftd>\n\u003Ctd>Apache 2.0\u003C\u002Ftd>\n\u003Ctd>\u003Ca href=\"\u002Fmodels\u002Fqwen3-8-27b-uncensored\u002F\">Qwen3.8 variants\u003C\u002Fa>\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>gpt-oss-20b\u003C\u002Ftd>\n\u003Ctd>Apache 2.0\u003C\u002Ftd>\n\u003Ctd>—\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Llama 3.3 70B Instruct\u003C\u002Ftd>\n\u003Ctd>Llama 3.3 Community License\u003C\u002Ftd>\n\u003Ctd>\u003Ca href=\"\u002Fmodels\u002Fllama-3-3-70b-abliterated\u002F\">Llama 3.3 70B abliterated\u003C\u002Fa>\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Llama 3.1 models\u003C\u002Ftd>\n\u003Ctd>Llama 3.1 Community License\u003C\u002Ftd>\n\u003Ctd>—\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Gemma 3 27B\u003C\u002Ftd>\n\u003Ctd>Gemma Terms of Use\u003C\u002Ftd>\n\u003Ctd>—\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftbody>\n\u003C\u002Ftable>\n\u003C\u002Fdiv>\n\u003Cp>Llama 3.3 70B Instruct and Gemma 3 27B are also gated on Hugging Face, so downloading them requires accepting the terms and waiting for manual approval.\u003C\u002Fp>\n\u003Ch2 id=\"comparison-using-the-model\">Comparison: using the model\u003C\u002Fh2>\n\u003Cdiv class=\"table-wrap\" role=\"region\" aria-label=\"Scrollable table\" tabindex=\"0\">\n\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth>Question\u003C\u002Fth>\n\u003Cth>MIT\u003C\u002Fth>\n\u003Cth>Apache 2.0\u003C\u002Fth>\n\u003Cth>Llama 3.3 \u002F 3.1 Community License\u003C\u002Fth>\n\u003Cth>Gemma Terms of Use\u003C\u002Fth>\n\u003C\u002Ftr>\n\u003C\u002Fthead>\n\u003Ctbody>\n\u003Ctr>\n\u003Ctd>Commercial use\u003C\u002Ftd>\n\u003Ctd>Yes; the grant includes the right to “sell copies”\u003C\u002Ftd>\n\u003Ctd>Yes\u003C\u002Ftd>\n\u003Ctd>Yes, below the user threshold\u003C\u002Ftd>\n\u003Ctd>Yes, within the terms\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>User-count threshold\u003C\u002Ftd>\n\u003Ctd>None\u003C\u002Ftd>\n\u003Ctd>None\u003C\u002Ftd>\n\u003Ctd>Above 700 million monthly active users on the version’s release date (6 December 2024 for 3.3, 23 July 2024 for 3.1), you must request a license from Meta\u003C\u002Ftd>\n\u003Ctd>None\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Use policy\u003C\u002Ftd>\n\u003Ctd>None\u003C\u002Ftd>\n\u003Ctd>None\u003C\u002Ftd>\n\u003Ctd>Version-specific Acceptable Use Policy, “incorporated by reference”\u003C\u002Ftd>\n\u003Ctd>Gemma Prohibited Use Policy, “incorporated by reference”\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Licensor’s extra powers\u003C\u002Ftd>\n\u003Ctd>None\u003C\u002Ftd>\n\u003Ctd>Patent license ends if you bring a patent suit claiming the work infringes\u003C\u002Ftd>\n\u003Ctd>All licenses end if you sue Meta or anyone else alleging the materials or outputs infringe your rights\u003C\u002Ftd>\n\u003Ctd>Google “reserves the right to restrict (remotely or otherwise) usage” it reasonably believes violates the terms\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Outputs\u003C\u002Ftd>\n\u003Ctd>Not addressed\u003C\u002Ftd>\n\u003Ctd>Not addressed\u003C\u002Ftd>\n\u003Ctd>No ownership clause; outputs appear in the naming rule and the AUP\u003C\u002Ftd>\n\u003Ctd>“Google claims no rights in Outputs you generate using Gemma”\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftbody>\n\u003C\u002Ftable>\n\u003C\u002Fdiv>\n\u003Ch2 id=\"comparison-redistributing-and-naming\">Comparison: redistributing and naming\u003C\u002Fh2>\n\u003Cdiv class=\"table-wrap\" role=\"region\" aria-label=\"Scrollable table\" tabindex=\"0\">\n\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth>Duty when you share weights\u003C\u002Fth>\n\u003Cth>MIT\u003C\u002Fth>\n\u003Cth>Apache 2.0\u003C\u002Fth>\n\u003Cth>Llama 3.3 \u002F 3.1 Community License\u003C\u002Fth>\n\u003Cth>Gemma Terms of Use\u003C\u002Fth>\n\u003C\u002Ftr>\n\u003C\u002Fthead>\n\u003Ctbody>\n\u003Ctr>\n\u003Ctd>Include the license\u003C\u002Ftd>\n\u003Ctd>Include the copyright and permission notice\u003C\u002Ftd>\n\u003Ctd>Give recipients “a copy of this License”\u003C\u002Ftd>\n\u003Ctd>“Provide a copy of this Agreement”\u003C\u002Ftd>\n\u003Ctd>Provide “a copy of this Agreement”\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Attribution\u003C\u002Ftd>\n\u003Ctd>The same notice\u003C\u002Ftd>\n\u003Ctd>Keep copyright, patent, trademark and attribution notices; pass on any NOTICE file’s attributions\u003C\u002Ftd>\n\u003Ctd>A Notice file with Meta’s attribution text, and “Built with Llama” displayed prominently on a related website, interface or documentation\u003C\u002Ftd>\n\u003Ctd>A Notice file stating that Gemma “is provided under and subject to the Gemma Terms of Use”\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Mark your changes\u003C\u002Ftd>\n\u003Ctd>Not required\u003C\u002Ftd>\n\u003Ctd>Modified files must “carry prominent notices”\u003C\u002Ftd>\n\u003Ctd>Not required by the text\u003C\u002Ftd>\n\u003Ctd>Modified files must “carry prominent notices”\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Naming\u003C\u002Ftd>\n\u003Ctd>No rule\u003C\u002Ftd>\n\u003Ctd>No trademark rights beyond describing the origin of the work\u003C\u002Ftd>\n\u003Ctd>“Llama” at the beginning of the name of any model built, trained or improved with Llama materials or outputs that you distribute or make available\u003C\u002Ftd>\n\u003Ctd>No trademark rights; you must not suggest endorsement by Google\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Pass on restrictions\u003C\u002Ftd>\n\u003Ctd>No\u003C\u002Ftd>\n\u003Ctd>No\u003C\u002Ftd>\n\u003Ctd>The agreement travels with the materials, and the AUP binds use\u003C\u002Ftd>\n\u003Ctd>The use restrictions must appear “as an enforceable provision” in your terms with recipients\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Add your own terms\u003C\u002Ftd>\n\u003Ctd>Allowed (the grant includes “sublicense”)\u003C\u002Ftd>\n\u003Ctd>Allowed for your modifications, if you meet the license’s conditions for the original\u003C\u002Ftd>\n\u003Ctd>Not addressed; as between you and Meta, you own your derivatives\u003C\u002Ftd>\n\u003Ctd>Allowed if they do not conflict with the Gemma terms\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Hosted service\u003C\u002Ftd>\n\u003Ctd>Not addressed\u003C\u002Ftd>\n\u003Ctd>Not addressed\u003C\u002Ftd>\n\u003Ctd>Duties apply to anything you “distribute or make available”\u003C\u002Ftd>\n\u003Ctd>Counts as “Distribution”, though no Notice file is required for a hosted service\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftbody>\n\u003C\u002Ftable>\n\u003C\u002Fdiv>\n\u003Ch2 id=\"notes-on-each-license\">Notes on each license\u003C\u002Fh2>\n\u003Ch3 id=\"mit-glm-5-3-flash-and-glm-4-7-flash\">MIT: GLM-5.3-Flash and GLM-4.7-Flash\u003C\u002Fh3>\n\u003Cp>The \u003Ca href=\"https:\u002F\u002Fhuggingface.co\u002Fzai-org\u002FGLM-5.3-Flash\u002Fblob\u002Fmain\u002FLICENSE\" rel=\"noopener\" class=\"ext\">GLM-5.3-Flash LICENSE file\u003C\u002Fa> is the standard MIT text with “Copyright © 2026 Z.AI Co., Ltd”. It grants the rights “to use, copy, modify, merge, publish, distribute, sublicense, and\u002For sell copies” on one condition: “The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.” It has no use policy, no naming rule and no explicit patent clause.\u003C\u002Fp>\n\u003Ch3 id=\"apache-2-0-gemma-4-qwen3-8-and-gpt-oss-20b\">Apache 2.0: Gemma 4, Qwen3.8 and gpt-oss-20b\u003C\u002Fh3>\n\u003Cp>\u003Ca href=\"https:\u002F\u002Fwww.apache.org\u002Flicenses\u002FLICENSE-2.0\" rel=\"noopener\" class=\"ext\">Apache 2.0\u003C\u002Fa> adds more process than MIT. When you redistribute, Section 4 requires you to give recipients the license, mark modified files, keep existing notices and carry forward any NOTICE file. Section 3 grants a patent license that ends if you sue claiming the work infringes a patent. Section 6 grants no trademark rights beyond “reasonable and customary use in describing the origin of the Work”.\u003C\u002Fp>\n\u003Cp>Google publishes the \u003Ca href=\"https:\u002F\u002Fai.google.dev\u002Fgemma\u002Fdocs\u002Fgemma_4_license\" rel=\"noopener\" class=\"ext\">Gemma 4 license\u003C\u002Fa> as the Apache License 2.0 text. Gemma 4 does not fall under the Gemma Terms of Use.\u003C\u002Fp>\n\u003Ch3 id=\"llama-3-3-community-license\">Llama 3.3 Community License\u003C\u002Fh3>\n\u003Cp>The \u003Ca href=\"https:\u002F\u002Fwww.llama.com\u002Fllama3_3\u002Flicense\u002F\" rel=\"noopener\" class=\"ext\">Llama 3.3 license\u003C\u002Fa> grants a “non-exclusive, worldwide, non-transferable and royalty-free limited license” to use, copy, modify and distribute. In return, Section 1.b requires a copy of the agreement, a “Built with Llama” display, the Notice text “Llama 3.3 is licensed under the Llama 3.3 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved.” and the naming rule. Section 2 sets the 700-million-user threshold. Section 5(a) grants use of the “Llama” mark only as needed for the naming rule.\u003C\u002Fp>\n\u003Cp>The \u003Ca href=\"https:\u002F\u002Fwww.llama.com\u002Fllama3_3\u002Fuse-policy\u002F\" rel=\"noopener\" class=\"ext\">Llama 3.3 Acceptable Use Policy\u003C\u002Fa> has a clause that matters for refusal-reduced models. Item 1 prohibits using Llama 3.3 to “Violate the law or others’ rights”, and its sub-item h names any action “to intentionally circumvent or remove usage restrictions or other safety measures, or to enable functionality disabled by Meta”. Whether making or using an \u003Ca href=\"\u002Flearn\u002Fwhat-is-abliteration\u002F\">abliterated\u003C\u002Fa> Llama 3.3 falls under that clause is a legal question this page does not answer. It is a reason to read the policy with your own use case in mind.\u003C\u002Fp>\n\u003Cp>The policy also withholds the Section 1(a) rights “with respect to any multimodal models included in Llama 3.3” from individuals and companies based in the European Union, though not from end users of products that incorporate them. Llama 3.3 70B Instruct accepts text only.\u003C\u002Fp>\n\u003Ch3 id=\"llama-3-1-community-license\">Llama 3.1 Community License\u003C\u002Fh3>\n\u003Cp>Apart from the version number and the policy it links to, the \u003Ca href=\"https:\u002F\u002Fwww.llama.com\u002Fllama3_1\u002Flicense\u002F\" rel=\"noopener\" class=\"ext\">Llama 3.1 license\u003C\u002Fa> text from Section 1 to Section 6 matched the 3.3 text word for word when we compared them. The \u003Ca href=\"https:\u002F\u002Fwww.llama.com\u002Fllama3_1\u002Fuse-policy\u002F\" rel=\"noopener\" class=\"ext\">Llama 3.1 Acceptable Use Policy\u003C\u002Fa> is shorter. As published on 14 November 2026, it did not contain the 3.3 policy’s clause about circumventing safety measures. Both policies say the most recent copy is on Meta’s site, so check the current text rather than relying on this summary.\u003C\u002Fp>\n\u003Ch3 id=\"gemma-terms-of-use-gemma-3\">Gemma Terms of Use: Gemma 3\u003C\u002Fh3>\n\u003Cp>The \u003Ca href=\"https:\u002F\u002Fai.google.dev\u002Fgemma\u002Fterms\" rel=\"noopener\" class=\"ext\">Gemma Terms of Use\u003C\u002Fa> apply to the models listed in their appendix: Gemma 1, 1.1, 2, 3 and 3n. Two definitions reach further than in other licenses. “Distribution” includes making Gemma available “as a hosted service via API, web access, or any other electronic or remote means”. “Model Derivatives” include models trained on Gemma’s synthetic outputs “to cause that model to perform similarly to Gemma”, though outputs themselves are not derivatives.\u003C\u002Fp>\n\u003Cp>The \u003Ca href=\"https:\u002F\u002Fai.google.dev\u002Fgemma\u002Fprohibited_use_policy\" rel=\"noopener\" class=\"ext\">Prohibited Use Policy\u003C\u002Fa> lists “Attempts to override or circumvent safety filters” under dangerous, illegal or malicious activities. As with the Llama clause, how that applies to a refusal-reduced Gemma 3 variant is a question for your own legal review.\u003C\u002Fp>\n\u003Ch2 id=\"what-changes-with-a-third-party-variant\">What changes with a third-party variant\u003C\u002Fh2>\n\u003Cp>A refusal-reduced variant is a modification of its base model. Abliteration edits the weights, a fine-tune trains them further, and a GGUF quantization converts them. The base license does not disappear in any of these steps.\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>Using a variant\u003C\u002Fstrong> brings in the base license’s use terms: the AUP or Prohibited Use Policy, the user threshold, and the law.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Redistributing one\u003C\u002Fstrong>, whether by uploading it, quantizing and re-uploading it, or serving it to others, brings in the redistribution duties in the table above.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Permissive licenses let the variant author add terms\u003C\u002Fstrong>, but only for their own changes. Under Apache 2.0, they may set different terms for their modifications provided the original work’s conditions are still met. The base notices stay.\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch3 id=\"when-variant-metadata-disagrees-with-the-base\">When variant metadata disagrees with the base\u003C\u002Fh3>\n\u003Cp>A Hugging Face license tag is a one-word field filled in by whoever uploads the model. It is not a license file, and it can be wrong.\u003C\u002Fp>\n\u003Cp>One example: the \u003Ca href=\"https:\u002F\u002Fhuggingface.co\u002FHauhauCS\u002FGemma4-26B-A4B-QAT-Uncensored-HauhauCS-Balanced-MTP\" rel=\"noopener\" class=\"ext\">HauhauCS Gemma 4 26B A4B QAT variant\u003C\u002Fa> names \u003Ccode>google\u002Fgemma-4-26B-A4B-it\u003C\u002Fcode> as its base model and carries the license tag \u003Ccode>gemma\u003C\u002Fcode>, which refers to the Gemma Terms of Use. Google publishes Gemma 4 under Apache 2.0, and the Gemma Terms of Use themselves say “For Gemma 4 terms, see the Gemma 4 license”. The card does not explain the tag.\u003C\u002Fp>\n\u003Cp>When a variant’s metadata and its base disagree:\u003C\u002Fp>\n\u003Col>\n\u003Cli>Follow \u003Ccode>base_model\u003C\u002Fcode> to the base repository and read its license file.\u003C\u002Fli>\n\u003Cli>Look for a separate license file in the variant’s repository. A tag alone does not state terms.\u003C\u002Fli>\n\u003Cli>Treat the disagreement as a reason to check, not as a new license. The base model’s license governs what the variant may be used for.\u003C\u002Fli>\n\u003Cli>If it matters for your use, ask the variant’s author and record the answer.\u003C\u002Fli>\n\u003C\u002Fol>\n\u003Cp>For a walk-through of reading a variant’s card, see \u003Ca href=\"\u002Flearn\u002Funcensored-vs-abliterated\u002F\">uncensored vs abliterated\u003C\u002Fa>. For the difference between a permissive license and an open source model, see \u003Ca href=\"\u002Flearn\u002Fopen-weight-vs-open-source\u002F\">open-weight vs open source\u003C\u002Fa>.\u003C\u002Fp>\n\u003Ch2 id=\"a-checklist-before-you-ship\">A checklist before you ship\u003C\u002Fh2>\n\u003Col>\n\u003Cli>Identify the base model and its version. Llama 3.1 and 3.3 have different use policies.\u003C\u002Fli>\n\u003Cli>Read the base license file and every policy it incorporates by reference.\u003C\u002Fli>\n\u003Cli>Check your use against the use policy, and your company’s size against any threshold.\u003C\u002Fli>\n\u003Cli>If you redistribute, include the license, the required notices and any required display text.\u003C\u002Fli>\n\u003Cli>If you publish a new model built from Llama 3.3 or 3.1, start its name with “Llama”.\u003C\u002Fli>\n\u003Cli>If you serve a Gemma 3 model to others, include the use restrictions in your terms.\u003C\u002Fli>\n\u003Cli>Save dated copies of the license and policy versions you relied on.\u003C\u002Fli>\n\u003C\u002Fol>\n\u003Cp>None of this replaces legal advice. It shows where the obligations are written, so you can ask a lawyer precise questions.\u003C\u002Fp>\n",[17,21,24,27,30,34,37,40,43,46,49,52],{"id":18,"text":19,"level":20},"which-license-covers-which-model","Which license covers which model",2,{"id":22,"text":23,"level":20},"comparison-using-the-model","Comparison: using the model",{"id":25,"text":26,"level":20},"comparison-redistributing-and-naming","Comparison: redistributing and naming",{"id":28,"text":29,"level":20},"notes-on-each-license","Notes on each license",{"id":31,"text":32,"level":33},"mit-glm-5-3-flash-and-glm-4-7-flash","MIT: GLM-5.3-Flash and GLM-4.7-Flash",3,{"id":35,"text":36,"level":33},"apache-2-0-gemma-4-qwen3-8-and-gpt-oss-20b","Apache 2.0: Gemma 4, Qwen3.8 and gpt-oss-20b",{"id":38,"text":39,"level":33},"llama-3-3-community-license","Llama 3.3 Community License",{"id":41,"text":42,"level":33},"llama-3-1-community-license","Llama 3.1 Community License",{"id":44,"text":45,"level":33},"gemma-terms-of-use-gemma-3","Gemma Terms of Use: Gemma 3",{"id":47,"text":48,"level":20},"what-changes-with-a-third-party-variant","What changes with a third-party variant",{"id":50,"text":51,"level":33},"when-variant-metadata-disagrees-with-the-base","When variant metadata disagrees with the base",{"id":53,"text":54,"level":20},"a-checklist-before-you-ship","A checklist before you ship",1694,[57,61,65,69,72,75,78,82,85,88],{"title":58,"publisher":59,"url":60},"Apache License, Version 2.0","Apache Software Foundation","https:\u002F\u002Fwww.apache.org\u002Flicenses\u002FLICENSE-2.0",{"title":62,"publisher":63,"url":64},"GLM-5.3-Flash LICENSE (MIT)","Hugging Face (zai-org)","https:\u002F\u002Fhuggingface.co\u002Fzai-org\u002FGLM-5.3-Flash\u002Fblob\u002Fmain\u002FLICENSE",{"title":66,"publisher":67,"url":68},"Llama 3.3 Community License Agreement","Meta","https:\u002F\u002Fwww.llama.com\u002Fllama3_3\u002Flicense\u002F",{"title":70,"publisher":67,"url":71},"Llama 3.3 Acceptable Use Policy","https:\u002F\u002Fwww.llama.com\u002Fllama3_3\u002Fuse-policy\u002F",{"title":73,"publisher":67,"url":74},"Llama 3.1 Community License Agreement","https:\u002F\u002Fwww.llama.com\u002Fllama3_1\u002Flicense\u002F",{"title":76,"publisher":67,"url":77},"Llama 3.1 Acceptable Use Policy","https:\u002F\u002Fwww.llama.com\u002Fllama3_1\u002Fuse-policy\u002F",{"title":79,"publisher":80,"url":81},"Gemma Terms of Use","Google AI for Developers","https:\u002F\u002Fai.google.dev\u002Fgemma\u002Fterms",{"title":83,"publisher":80,"url":84},"Gemma Prohibited Use Policy","https:\u002F\u002Fai.google.dev\u002Fgemma\u002Fprohibited_use_policy",{"title":86,"publisher":80,"url":87},"Gemma 4 license (Apache License 2.0)","https:\u002F\u002Fai.google.dev\u002Fgemma\u002Fdocs\u002Fgemma_4_license",{"title":89,"publisher":90,"url":91},"Gemma4-26B-A4B-QAT-Uncensored-HauhauCS-Balanced-MTP model card","Hugging Face (HauhauCS)","https:\u002F\u002Fhuggingface.co\u002FHauhauCS\u002FGemma4-26B-A4B-QAT-Uncensored-HauhauCS-Balanced-MTP",[93,96,99,102],{"q":94,"a":95},"Can I use an abliterated Llama 3.3 commercially?","The Llama 3.3 license allows commercial use below its 700-million-user threshold, and an abliterated variant carries the same license. Its Acceptable Use Policy still applies, including a clause on circumventing safety measures, so read it against your use case and get legal advice if the stakes are real.",{"q":97,"a":98},"Do I have to put 'Llama' in my model's name?","If you use Llama 3.3 or 3.1 materials, or their outputs, to create, train or improve a model that you distribute or make available, both licenses say its name must begin with 'Llama'.",{"q":100,"a":101},"Is Gemma 4 under the Gemma Terms of Use?","No. Google's Gemma Terms of Use cover Gemma 1, 1.1, 2, 3 and 3n and point to a separate Gemma 4 license, which is the Apache License 2.0.",{"q":103,"a":104},"Does the MIT license require attribution?","Yes, in a minimal form: the copyright notice and the permission notice must be included in all copies or substantial portions. It has no naming rule and no use policy.",[106,110,114,118],{"path":107,"title":108,"description":109},"\u002Flearn\u002Fopen-weight-vs-open-source\u002F","Open-weight vs open-source AI","Open-weight means you can download the weights. Open source AI means meeting the OSI definition. What MIT, Apache 2.0, Llama and Gemma terms actually allow.",{"path":111,"title":112,"description":113},"\u002Flearn\u002Funcensored-vs-abliterated\u002F","Uncensored vs abliterated LLMs","Abliterated names one weight-editing method; uncensored is a looser label. Compare data, compute, side effects and checks, and learn to read a model card.",{"path":115,"title":116,"description":117},"\u002Fmodels\u002F","Model catalogue","Every catalogued model with sources, filters and sorting.",{"path":119,"title":120,"description":121},"\u002Fapi\u002Fproviders\u002F","API providers","Provider profiles with published features and limits.",null,[124,125,126,127,128,129,111,107],"\u002Fmodels\u002Fglm-5-3-flash-uncensored\u002F","\u002Fmodels\u002Fglm-4-7-flash-uncensored\u002F","\u002Fmodels\u002Fgemma-4-26b-a4b-uncensored\u002F","\u002Fmodels\u002Fqwen3-8-27b-uncensored\u002F","\u002Fmodels\u002Fllama-3-3-70b-abliterated\u002F","\u002Flearn\u002Fwhat-is-abliteration\u002F",1794622336927]