{"models":[{"name":"1.58-bit large language model","slug":"1-58-bit-large-language-model","tagline":"A model whose weights hold only three values: -1, 0 and +1.","kind":["Research","General"],"access":["Open weights","Research only"],"arch":["Dense decoder"],"modality":["Text"],"lang":[],"params":{"step":"1 of 5","label":"Compact","span":"1 to 10 billion","value":2,"unit":"billion parameters"},"released":"2024","main_country":null,"region":null,"developers_of":["Microsoft"],"photo":null,"url":"https://everyllm.cpn.cleartrust.cc/m/1-58-bit-large-language-model"},{"name":"Mistral Large 3","slug":"mistral-large-3","tagline":"A 675B-parameter open-weight mixture-of-experts multimodal model from Mistral AI","kind":["General","Multimodal"],"access":["Open weights","API only"],"arch":["Mixture of experts"],"modality":["Text","Images in","Tools and agents"],"lang":["Many languages"],"params":{"step":"4 of 5","label":"Very large","span":"300 billion to 1 trillion","value":675,"unit":"billion parameters"},"released":null,"main_country":"France","region":"Europe","developers_of":["Mistral AI","NVIDIA","vLLM","Red Hat"],"photo":null,"url":"https://everyllm.cpn.cleartrust.cc/m/mistral-large-3"},{"name":"Mistral Large","slug":"mistral-large","tagline":"Mistral AI's flagship multilingual text model, released in 2024 and now deprecated.","kind":["General"],"access":["API only","Retired"],"arch":["Dense decoder"],"modality":["Text"],"lang":["European languages","English first"],"params":null,"released":"2024","main_country":"France","region":"Europe","developers_of":["Mistral AI"],"photo":null,"url":"https://everyllm.cpn.cleartrust.cc/m/mistral-large"},{"name":"Hunyuan-Large","slug":"hunyuan-large","tagline":"Open-source mixture of experts model with 389 billion parameters","kind":["General","Multimodal"],"access":["Open weights"],"arch":["Mixture of experts"],"modality":["Text"],"lang":["Many languages","English first","Chinese"],"params":{"step":"4 of 5","label":"Very large","span":"300 billion to 1 trillion","value":389,"unit":"billion parameters"},"released":"November 2024","main_country":"China","region":"East Asia","developers_of":["Tencent"],"photo":null,"url":"https://everyllm.cpn.cleartrust.cc/m/hunyuan-large"},{"name":"Trinity Large","slug":"trinity-large","tagline":"398B-parameter sparse MoE model family by Arcee AI","kind":["General","Reasoning"],"access":["Open weights","API only"],"arch":["Mixture of experts"],"modality":["Text","Tools and agents"],"lang":["Many languages"],"params":{"step":"4 of 5","label":"Very large","span":"300 billion to 1 trillion","value":398,"unit":"billion parameters"},"released":"2026","main_country":"United States","region":"North America","developers_of":["Arcee AI"],"photo":null,"url":"https://everyllm.cpn.cleartrust.cc/m/trinity-large"},{"name":"GPT-4.1","slug":"gpt-4-1","tagline":"Multimodal large language model released by OpenAI in April 2025.","kind":["General","Multimodal"],"access":["API only"],"arch":["Dense decoder"],"modality":["Text","Images in"],"lang":["Many languages"],"params":null,"released":"14 April 2025","main_country":"United States","region":"North America","developers_of":["OpenAI"],"photo":null,"url":"https://everyllm.cpn.cleartrust.cc/m/gpt-4-1"},{"name":"Gemini","slug":"gemini","tagline":"Google DeepMind's multimodal LLM family, from Nano to Ultra","kind":["General","Multimodal"],"access":["API only","Fully open"],"arch":["Dense decoder","Mixture of experts"],"modality":["Text","Images in","Speech and audio","Video in"],"lang":["Many languages","English first"],"params":null,"released":"6 December 2023","main_country":"United States","region":"North America","developers_of":["Google DeepMind","Google AI"],"photo":null,"url":"https://everyllm.cpn.cleartrust.cc/m/gemini"},{"name":"Wu Dao","slug":"wu-dao","tagline":"BAAI's trillion-parameter multimodal model family, announced 2021","kind":["Multimodal","General"],"access":["Research only"],"arch":["Mixture of experts"],"modality":["Text","Images in","Images out"],"lang":["Chinese","English first","Many languages"],"params":{"step":"5 of 5","label":"Trillion-scale","span":"1 trillion and more","value":1750,"unit":"billion parameters"},"released":"11 January 2021","main_country":"China","region":"East Asia","developers_of":["Beijing Academy of Artificial Intelligence"],"photo":null,"url":"https://everyllm.cpn.cleartrust.cc/m/wu-dao"},{"name":"GPT-5","slug":"gpt-5","tagline":"Multimodal large language model released by OpenAI in August 2025.","kind":["Multimodal","Assistant"],"access":["API only"],"arch":["Dense decoder"],"modality":["Text","Images in","Tools and agents"],"lang":["Many languages"],"params":null,"released":"7 August 2025","main_country":"United States","region":"North America","developers_of":["OpenAI"],"photo":null,"url":"https://everyllm.cpn.cleartrust.cc/m/gpt-5"},{"name":"XLNet","slug":"xlnet","tagline":"Permutation-based autoregressive transformer model created by Google AI","kind":["Research","General"],"access":["Open weights"],"arch":["Dense decoder"],"modality":["Text"],"lang":["English first"],"params":{"step":"0 of 5","label":"Small","span":"under 1 billion","value":0.34,"unit":"billion parameters"},"released":"19 June 2019","main_country":null,"region":null,"developers_of":["Google AI"],"photo":null,"url":"https://everyllm.cpn.cleartrust.cc/m/xlnet"}],"countries":[],"hubs":[],"concepts":[{"term":"Large language model","definition":"A neural network trained on very large amounts of text to predict the next token, and so to read and write language.","url":"https://everyllm.cpn.cleartrust.cc/concepts/large-language-model"}],"events":[],"published_by":"EveryLLM, a ClearTrust publication"}