Glossary
Apache 2.0
A permissive open-source license used by most open-weight model releases (Llama from 4 onward partial, Qwen, Mistral, DeepSeek, Falcon), allowing commercial use without acceptable-use restrictions.
A permissive license approved by the OSIgovernanceThe nonprofit that maintains the canonical Open Source Definition for software since 1998, and the OSAID definition for AI as of 2024. Open full entry . The operative grant: a perpetual, irrevocable, royalty-free copyright and patent license to use, modify, and distribute, including in commercial products and proprietary derivatives. The notable add over MIT or BSD: explicit patent grant with a retaliation clause.
For model weights, Apache 2.0 is the strongest “actually open” signal. It carries no acceptable-usegovernanceLicense or terms-of-service clauses that prohibit certain uses (weapons, surveillance, harassment, child sexual abuse material), common on open-weight licenses but rejected by the strict open-source definition. Open full entry clause, no MAUgovernanceA user-count metric used in restrictive open-weights licenses (notably Llama's Community License) to trigger a requirement to negotiate a separate commercial license at scale. Open full entry threshold, no field-of-usegovernanceLicense clauses that limit which industries or applications a model may be deployed in, restricting use to non-competitive, non-commercial, or non-government purposes. Open full entry restriction. A team can fine-tuningtrainingContinued training of a pretrained base model on a smaller, task-specific dataset to specialize its behavior without retraining from scratch. Open full entry the model, ship the result behind a proprietary product, and never report back to the original publisher.
In 2026 the meaningful Apache-2.0 open weightsweightsA model release that publishes the trained parameters under some downloadable license, distinct from "open source" which (per OSAID) also requires data and training-code openness. Open full entry families include MistralweightsA French open-weight model family from Mistral AI, released mostly under Apache 2.0 with strong performance per parameter and notable MoE variants (Mixtral, Mixtral 8x22B). Open full entry (most releases), QwenweightsAlibaba's open-weight model family, leading the multilingual and Chinese-language open-weight space, released under Apache 2.0 with sizes from 0.6B to 235B parameters. Open full entry (most releases), DeepSeekweightsA Chinese open-weight family known for the V3 MoE base model and the R1 reasoning model, both released under permissive licenses and unusually transparent in their training-cost reporting. Open full entry (most releases), and OLMo (AI2). Phi (Microsoft) ships under MIT, which is similarly permissive but lacks the explicit patent grant; Falcon (TII) ships under a TII Falcon License derived from Apache 2.0 with an acceptable-use addendum. LlamaweightsMeta's open-weight model family, the most widely deployed open release through 2024 to 2026, released under the source-available Community License with an MAU cap and acceptable-use clause. Open full entry is closer to source-available: the Community License has both a 700M MAUgovernanceA user-count metric used in restrictive open-weights licenses (notably Llama's Community License) to trigger a requirement to negotiate a separate commercial license at scale. Open full entry clause and an acceptable-usegovernanceLicense or terms-of-service clauses that prohibit certain uses (weapons, surveillance, harassment, child sexual abuse material), common on open-weight licenses but rejected by the strict open-source definition. Open full entry addendum, neither compatible with the OSIgovernanceThe nonprofit that maintains the canonical Open Source Definition for software since 1998, and the OSAID definition for AI as of 2024. Open full entry definition.