News
EveryLLM

Open weights, permissive licence

Weights to download under Apache 2.0, MIT or similar: Mistral 7B, Qwen, DeepSeek, Falcon. 272 models on EveryLLM so far.

Open-weights models with permissive licences publish their trained parameters for anyone to download, run, fine-tune and build into products, under terms such as Apache 2.0 or MIT that ask for little more than keeping the copyright notice. The training data and full training code are usually not released, which is why many people prefer the term open weights to open source. Mistral 7B, Qwen, DeepSeek, Phi and Granite are leading examples.

What the licence allows

Apache 2.0 and MIT allow commercial use, modification and redistribution without asking permission or paying fees, and Apache 2.0 also grants a patent licence from contributors. Mistral AI released Mistral 7B under Apache 2.0 in September 2023, Abu Dhabi's TII moved Falcon 40B to Apache 2.0 in 2023, and DeepSeek released R1 under the MIT licence in January 2025. OpenAI's gpt-oss models (August 2025) were its first open-weights release since GPT-2.

Why labs give weights away

Publishing weights builds a developer community, spreads a lab's tools and earns goodwill with researchers and governments. Users gain control: they can run the model on their own hardware, keep data private and inspect its behaviour. The debate is over risk. Supporters say openness speeds safety research and competition; critics worry that safeguards can be removed by fine-tuning once weights are public. Some labs now keep only their largest models closed.

Famous open weights models

Further reading

XLNet WorldXLNet Permutation-based autoregressive transformer model created by Google AIResearch 路 Open weights 路 2019 Apertus 馃嚚馃嚟 SwitzerlandApertus A Swiss public large language model released in 8B and 70B sizesGeneral 路 Open weights 路 2025 GPT-J WorldGPT-J A 6-billion-parameter open English text generator from EleutherAI, released in 2021.General 路 Open weights 路 2021 1.58-bit large language model World1.58-bit large language model A model whose weights hold only three values: -1, 0 and +1.Research 路 Open weights 路 2024 imgwavInkling WorldInkling An open-weights mixture-of-experts model handling text, images and audio inputGeneral 路 Open weights 路 2026 fnGLM 馃嚚馃嚦 ChinaGLM Open-weight Chinese model series used for chat and codingGeneral 路 Open weights 路 2021 GPT-4chan WorldGPT-4chan A GPT-J model fine-tuned on posts from 4chan's /pol/ board.General 路 Open weights 路 2022 IBM Granite 馃嚭馃嚫 United StatesIBM Granite IBM's decoder-only foundation models, from 3 to 34 billion parametersGeneral 路 Open weights 路 2023 Velvet AI 馃嚠馃嚬 ItalyVelvet AI Italian multilingual open-weight LLMs built for energy efficiencyGeneral 路 Open weights 路 2025 imgwavvidQwen 馃嚚馃嚦 ChinaQwen A family of large and small language models developed by Alibaba Cloud.General 路 Open weights 路 2023 wavWhisper WorldWhisper OpenAI's open-source model for turning speech into text and translating it.General 路 Open weights 路 2022 imgwavvidGemma 馃嚭馃嚫 United StatesGemma Google DeepMind's open family of Gemini-derived language modelsGeneral 路 Open weights 路 2024 imgfnPhi 馃嚭馃嚫 United StatesPhi Microsoft's family of small, open-weight language models that run on everyday devices.Small 路 Open weights 路 2023 imgHuawei PanGu 馃嚚馃嚦 ChinaHuawei PanGu Huawei's multimodal large model family for industry, from 1B to trillion-scaleMultimodal 路 Open weights 路 2023 Jais 馃嚘馃嚜 United Arab EmiratesJais An open Arabic and English large language model from the UAEGeneral 路 Open weights 路 2023 GPT-1 WorldGPT-1 OpenAI's first GPT model, a 117M-parameter decoder-only transformer from 2018.Research 路 Open weights 路 2018 Minerva 馃嚠馃嚬 ItalyMinerva An Italian-first large language model trained from scratch at Sapienza University of Rome.General 路 Open weights 路 2024 GPT-2 馃嚭馃嚫 United StatesGPT-2 OpenAI's 2019 text model that predicts the next word in a sequenceGeneral 路 Open weights 路 2019 Phi-2 馃嚭馃嚫 United StatesPhi-2 A 2.7 billion parameter Microsoft research model for QA, chat and codeResearch 路 Open weights MT5 馃嚭馃嚫 United StatesMT5 A multilingual text-to-text transformer pretrained on 101 languagesGeneral 路 Open weights 路 2020 Mixtral 8x22B 馃嚝馃嚪 FranceMixtral 8x22B Sparse mixture-of-experts open model with 39B active parameters out of 141BGeneral 路 Open weights imgwavvidGemma 4 馃嚭馃嚫 United StatesGemma 4 Google DeepMind's open multimodal models in five sizes, from phones to serversGeneral 路 Open weights imgvidfnGemma 3 馃嚭馃嚫 United StatesGemma 3 Lightweight multimodal open models from 1B to 27B parametersGeneral 路 Open weights 路 2025 Gemma 2 馃嚭馃嚫 United StatesGemma 2 Open lightweight language models from Google DeepMind, 2B to 27B parametersGeneral 路 Open weights 路 2024 fnSolar Open 2 馃嚢馃嚪 South KoreaSolar Open 2 250B parameter open-weight model for agentic tasksGeneral 路 Open weights 路 2026 RecurrentGemma 馃嚭馃嚫 United StatesRecurrentGemma Open language models using Google's Griffin recurrent-plus-attention architectureGeneral 路 Open weights 路 2024 fnGranite 4.2 馃嚭馃嚫 United StatesGranite 4.2 IBM's open reasoning language model family, released August 2026Reasoning 路 Open weights 路 2026 Llama 2 馃嚭馃嚫 United StatesLlama 2 Meta's open family of pretrained and chat-tuned language models, 7B to 70BGeneral 路 Open weights 路 2023 Phi-4 馃嚭馃嚫 United StatesPhi-4 A 14B Microsoft Research model trained heavily on synthetic data for reasoning.General 路 Open weights 路 2024 fnBlenderBot 3 WorldBlenderBot 3 A 175B open-domain chatbot with internet access and long-term memoryGeneral 路 Open weights 路 2022 Aya 101 WorldAya 101 A 13-billion-parameter multilingual model that follows instructions in 101 languagesGeneral 路 Open weights 路 2024 DeepSeek-V2 馃嚚馃嚦 ChinaDeepSeek-V2 Efficient mixture-of-experts language model with 236 billion parametersGeneral 路 Open weights 路 2024 Swallow 馃嚡馃嚨 JapanSwallow Japanese-enhanced Llama models by Tokyo TechGeneral 路 Open weights 路 2024 ALLaM 馃嚫馃嚘 Saudi ArabiaALLaM Bilingual Arabic and English language model seriesGeneral 路 Open weights 路 2024 Falcon2-11B 馃嚘馃嚜 United Arab EmiratesFalcon2-11B An 11 billion parameter causal decoder model by TII.General 路 Open weights 路 2024 Sarvam-M 馃嚠馃嚦 IndiaSarvam-M Hybrid reasoning model for Indian languages and EnglishGeneral 路 Open weights 路 2025 fnGpt-oss-20b 馃嚭馃嚫 United StatesGpt-oss-20b OpenAI's smaller open-weight reasoning model for local and low-latency useReasoning 路 Open weights 路 2025 Falcon-7B 馃嚘馃嚜 United Arab EmiratesFalcon-7B Open 7-billion parameter decoder model by TIIGeneral 路 Open weights 路 2023 MiMo-V2-Flash 馃嚚馃嚦 ChinaMiMo-V2-Flash High-speed MoE language model for reasoning and agentsGeneral 路 Open weights imgYi-34B 馃嚚馃嚦 ChinaYi-34B A 34-billion parameter bilingual language model by 01.AIGeneral 路 Open weights 路 2023 Falcon 180B 馃嚘馃嚜 United Arab EmiratesFalcon 180B Open 180 billion parameter language model from TIIGeneral 路 Open weights 路 2023 Project Indus 馃嚠馃嚦 IndiaProject Indus Open-source language model for Hindi and dialectsGeneral 路 Open weights ERNIE 5.5 Flash 馃嚚馃嚦 ChinaERNIE 5.5 Flash Efficient sparse MoE language model with 280B total parametersGeneral 路 Open weights Sarashina2 馃嚡馃嚨 JapanSarashina2 Open-weight Japanese and English language models by SB IntuitionsGeneral 路 Open weights SEA-LION 馃嚫馃嚞 SingaporeSEA-LION Multilingual language models for Southeast AsiaGeneral 路 Open weights 路 2025 SOLAR 10.7B 馃嚢馃嚪 South KoreaSOLAR 10.7B An 11-billion parameter language model using depth up-scaling.General 路 Open weights 路 2023 RakutenAI-7B 馃嚡馃嚨 JapanRakutenAI-7B Japanese-focused open language model based on MistralGeneral 路 Open weights 路 2024 PLLuM 馃嚨馃嚤 PolandPLLuM Polish-focused large language model familyGeneral 路 Open weights 路 2024

The weekly digest

Every Monday: the model of the day, new pages on EveryLLM, and the conferences and anniversaries of the week ahead.