News
ReleasesOVHcloud unveils Olala mini, its first in-house LLMChinaA 9B model beats a 1T model on FinQA, trained with Crusoe Managed RLReleasesThe internal metaphysics of an LLMReleasesMapping Language Model Internal Representations to Physical Laws and Scientific ReasoningResearchCan China keep rogue AI in check? Meet the start-up testing the top modelsFrontierMicrosoft aims to challenge OpenAI, Anthropic with cheaper, specialized AI modelsOpen modelsMicrosoft rolls out new first-party models and updates to agentic AI continuous improvementFrontierMicrosoft unveils AI models for coding, cybersecurity as it takes on Anthropic, OpenAIFrontierGoogle is making it easier to opt out of Gemini AI model training – here’s howFrontierOpenAI Unveils New Astra AI Model After Scrapping Prior ReleaseReleasesOVHcloud unveils Olala mini, its first in-house LLMChinaA 9B model beats a 1T model on FinQA, trained with Crusoe Managed RLReleasesThe internal metaphysics of an LLMReleasesMapping Language Model Internal Representations to Physical Laws and Scientific ReasoningResearchCan China keep rogue AI in check? Meet the start-up testing the top modelsFrontierMicrosoft aims to challenge OpenAI, Anthropic with cheaper, specialized AI modelsOpen modelsMicrosoft rolls out new first-party models and updates to agentic AI continuous improvementFrontierMicrosoft unveils AI models for coding, cybersecurity as it takes on Anthropic, OpenAIFrontierGoogle is making it easier to opt out of Gemini AI model training – here’s howFrontierOpenAI Unveils New Astra AI Model After Scrapping Prior Release
EveryLLM

Dense decoder transformers

The GPT design: one stack of transformer layers predicting the next token. 308 models on EveryLLM so far.

The dense decoder transformer is the design behind GPT and most chatbots. It is a single stack of identical transformer layers that reads the tokens so far and predicts the next one, with a mask that stops each position from looking ahead. Dense means every parameter is used for every token, unlike in a mixture of experts. GPT-3, PaLM, Llama 3, Chinchilla, Gemma, Phi and OLMo are dense decoders.

Inside each layer

Each layer has two parts. Self-attention lets every token weigh every earlier token and gather what it needs; a feed-forward network then transforms each token's vector on its own. The transformer was introduced by Google researchers in the June 2017 paper Attention Is All You Need, for translation, and OpenAI's GPT-1 (2018) kept only its decoder half. Llama 3.1 405B, released in July 2024, stacks 126 such layers.

Why it won

A decoder trains on any text without labels, because the next token is always the answer to predict, and it scales smoothly: scaling laws published by OpenAI in 2020 and DeepMind in 2022 showed the error falling predictably with more parameters, data and compute. Its weak point is cost. Attention grows with the square of the context length, and every parameter must be read for every token, which is why mixtures of experts became popular.

Famous dense decoder models

Still to be written: OLMo (language model).

Further reading

Galactica 🇺🇸 United StatesGalactica A language model trained on scientific text for research tasksDomain · Conditional licence · 2022 Apple Foundation Models 🇺🇸 United StatesApple Foundation Models Apple's on-device and server language models behind Apple IntelligenceGeneral · API only · 2024 imgMM1 WorldMM1 A family of multimodal language models up to 30B parameters, dense and MoE.Multimodal · Research only · 2024 Orca 2 🇺🇸 United StatesOrca 2 A small LLaMA-2-based model trained on synthetic data to reason.Small · Conditional licence · 2023 Grok-1.5 🇺🇸 United StatesGrok-1.5 xAI's long-context Grok model with stronger reasoning and coding skillsGeneral · API only · 2024 imgwavfnGrok 4 🇺🇸 United StatesGrok 4 xAI's flagship reasoning model with native tool use and live searchGeneral · API only · 2025 imgGrok-2 🇺🇸 United StatesGrok-2 xAI's 2024 chat and reasoning model, with a smaller sibling, Grok-2 miniGeneral · Conditional licence · 2024 Grok 4.1 🇺🇸 United StatesGrok 4.1 xAI's Grok 4.1, tuned for personality, emotional tone and fewer factual errorsGeneral · API only · 2025 fnCommand A WorldCommand A Cohere's enterprise-focused large language model for agents, search and tool useGeneral · Research only · 2025 Command R+ 🇨🇦 CanadaCommand R+ Cohere's large language model built for enterprise workloads and retrieval-augmented use.General · API only Phi-2 🇺🇸 United StatesPhi-2 A 2.7 billion parameter Microsoft research model for QA, chat and codeResearch · Open weights InstructGPT 🇺🇸 United StatesInstructGPT GPT-3 models fine-tuned with human feedback to follow instructionsGeneral · API only · 2022 fnWebGPT 🇺🇸 United StatesWebGPT A GPT-3 model fine-tuned to answer questions while browsing the webGeneral · Research only · 2021 Claude 1 🇺🇸 United StatesClaude 1 Anthropic's first AI assistant model, offered in two versions via chat and API.Assistant · API only Claude 2 🇺🇸 United StatesClaude 2 Anthropic's 2023 assistant model with a 100K-token window and stronger coding.General · API only · 2023 imgClaude 3 Opus 🇺🇸 United StatesClaude 3 Opus Anthropic's largest Claude 3 model, aimed at complex reasoning and analysisGeneral · API only imgClaude 3 Haiku 🇺🇸 United StatesClaude 3 Haiku Anthropic's fastest and lowest-cost model in the Claude 3 familyGeneral · API only · 2024 imgClaude 3.5 Haiku 🇺🇸 United StatesClaude 3.5 Haiku Anthropic's fast, low-latency model for coding, tools and user-facing productsGeneral · API only · 2024 imgfnClaude 3.7 Sonnet 🇺🇸 United StatesClaude 3.7 Sonnet Anthropic's hybrid reasoning model with a controllable thinking budgetGeneral · API only · 2025 fnClaude 4 🇺🇸 United StatesClaude 4 Anthropic's fourth-generation coding and reasoning models, Opus 4 and Sonnet 4General · API only · 2025 fnClaude Opus 4.1 🇺🇸 United StatesClaude Opus 4.1 Anthropic's upgraded Claude Opus 4 model for coding, agents and reasoningGeneral · API only · 2025 fnClaude Sonnet 4.5 🇺🇸 United StatesClaude Sonnet 4.5 Anthropic's coding and agentic Claude model, released in 2025.General · API only · 2025 fnClaude Haiku 4.5 🇺🇸 United StatesClaude Haiku 4.5 Anthropic's small, fast Claude model for coding, agents and real-time workGeneral · API only · 2025 imgfnClaude Opus 4.5 🇺🇸 United StatesClaude Opus 4.5 Anthropic's flagship Claude model for coding, agents and computer useGeneral · API only · 2025 imgfnClaude Sonnet 4.6 🇺🇸 United StatesClaude Sonnet 4.6 Anthropic's Sonnet-tier model for coding, computer use and long-context workGeneral · API only imgfnClaude Opus 5 🇺🇸 United StatesClaude Opus 5 Anthropic's Claude Opus model for coding, agents and knowledge workGeneral · API only fnClaude Opus 4.6 🇺🇸 United StatesClaude Opus 4.6 Anthropic's largest Claude model, built for long agentic coding and knowledge workGeneral · API only · 2026 imgwavvidGemini 1.0 🇺🇸 United StatesGemini 1.0 Google's first Gemini models: multimodal AI in Ultra, Pro and Nano sizesMultimodal · API only · 2023 imgwavvidGemini 3.1 Pro 🇺🇸 United StatesGemini 3.1 Pro Google DeepMind's multimodal reasoning model, the next step after Gemini 3 ProGeneral · API only · 2026 imgwavvidGemini 3.1 Flash-Lite 🇺🇸 United StatesGemini 3.1 Flash-Lite A fast, low-cost multimodal reasoning model for high-volume tasksMultimodal · API only · 2026 imgwavvidGemini 3.5 Flash 🇺🇸 United StatesGemini 3.5 Flash A multimodal reasoning model from Google DeepMind with adjustable thinking levelsGeneral · API only · 2026 Gemma 2 🇺🇸 United StatesGemma 2 Open lightweight language models from Google DeepMind, 2B to 27B parametersGeneral · Open weights · 2024 Gemma 🇺🇸 United StatesGemma Small open-weight language models from Google, in 2B and 7B sizesGeneral · Conditional licence · 2024 imgvidfnGemma 3 🇺🇸 United StatesGemma 3 Lightweight multimodal open models from 1B to 27B parametersGeneral · Open weights · 2025 imgwavvidGemma 3n 🇺🇸 United StatesGemma 3n Mobile-first open model family for on-device text, image and audio workMultimodal · Open weights · 2025 imgwavvidGemma 4 🇺🇸 United StatesGemma 4 Google DeepMind's open multimodal models in five sizes, from phones to serversGeneral · Open weights imgMedGemma 🇺🇸 United StatesMedGemma Medical vision-language models built on Gemma 3 for images and textMultimodal · Conditional licence · 2025 Med-PaLM 🇺🇸 United StatesMed-PaLM A medical-domain language model aligned from Flan-PaLM using instruction prompt tuning.Domain · Research only · 2022 Med-PaLM 2 🇺🇸 United StatesMed-PaLM 2 Google's medical question-answering model built on PaLM 2Domain · Research only · 2023 Minerva 🇺🇸 United StatesMinerva A language model trained further on technical content to solve maths and science problemsResearch · Research only · 2022 Llama 2 🇺🇸 United StatesLlama 2 Meta's open family of pretrained and chat-tuned language models, 7B to 70BGeneral · Open weights · 2023 Gopher 🇬🇧 United KingdomGopher A 280 billion parameter Transformer language model from DeepMind, studied for how scale changes ability.Research · Research only · 2021 fnLlama 3.1 405B 🇺🇸 United StatesLlama 3.1 405B Meta's largest openly available language model, released in 2024General · Conditional licence · 2024 imgLlama 3.2 🇺🇸 United StatesLlama 3.2 Small text models and vision models from Meta for edge devices and image tasksMultimodal · Open weights · 2024 Phi-4 🇺🇸 United StatesPhi-4 A 14B Microsoft Research model trained heavily on synthetic data for reasoning.General · Open weights · 2024 Code Llama 🇺🇸 United StatesCode Llama Meta's code-specialised family of Llama 2 models in four sizesCode · Conditional licence · 2023 Llama Guard WorldLlama Guard A Llama 2 based model that classifies prompts and replies for safety risksDomain · Open weights · 2023 fnBlenderBot 3 WorldBlenderBot 3 A 175B open-domain chatbot with internet access and long-term memoryGeneral · Open weights · 2022

The weekly digest

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