Large Language Models
(LLMs)

OpenAI's GPT model family and the ChatGPT assistant, also available through the OpenAI API and Codex.

Anthropic's AI assistant and model family — the Fable, Opus, Sonnet and Haiku tiers — built with a focus on safety, helpfulness and honesty.

Google's AI assistant and multimodal Gemini model family, the successor to Bard and PaLM 2.

xAI's Grok models and assistant, aimed at coding, agentic workflows and real-time information tasks.

A European AI platform offering open and commercial models, assistants and agents that enterprises can customize, fine-tune and deploy.

The AI lab behind the open-weight DeepSeek-V4 and DeepSeek-R1 models, with a chat app and an API.

Alibaba's Qwen models and chatbot, covering image and video understanding, document processing, web search and tool use.

Moonshot AI's Kimi assistant and models, built for agentic coding and knowledge work.

The Z.ai assistant, powered by the GLM model family, for building websites, writing code and handling long-horizon tasks.

A multimodal model company building AI-native products used by hundreds of millions of people.

Meta's Muse models, starting with Muse Spark, available through Meta AI and the Meta Model API. Muse Spark is Meta's first frontier model without open weights.

Meta's open-weight Llama model family. Meta ended the Llama line in 2026 and moved to its Muse models.

Microsoft's AI assistant for chat, search and getting work done across your day.

An AI-powered answer engine that searches the web and responds with cited sources.

Cohere's Command family of scalable language models for enterprise applications.

Amazon's family of foundation models and services, available through AWS.

IBM's family of open, trusted models for business, covering language, vision, speech and guardrail models.

Microsoft's efficient small language models for generative AI applications, available in Azure AI Foundry.

Google DeepMind's family of open models, built from the research behind Gemini.

OpenAI's open-weight language models, which can be downloaded and run on your own infrastructure.

NVIDIA's family of open models, released with open weights, training data and recipes, for building specialized reasoning agents.

Ai2's fully open language model, released together with the complete model flow.

A public leaderboard where people compare and vote on the best LLM, image and code models through real-world evaluation.

Independent benchmarks comparing AI models and API providers on quality, price, output speed and latency.

The open hub for exploring, comparing and downloading machine learning models, including most open-weight LLMs.

AI21 Labs' Jamba family — hybrid Mamba-Transformer models with a long context window, available as open weights and through an API.

The Technology Innovation Institute's Falcon model family from Abu Dhabi, released with open weights for research and commercial use.

Liquid Foundation Models (LFM) — compact, memory-efficient models designed to run on phones, laptops and edge devices.

Hugging Face's SmolLM family of small, fully open language models, with training data and recipes published openly.

Reka's multimodal models that understand text, images, audio and video, offered through an API and a chat assistant.

A single API and catalogue for hundreds of LLMs from many providers, with side-by-side pricing, context length and usage rankings.

A permissively licensed open-source reproduction of Meta AI's LLaMA 7B, trained on the RedPajama dataset — ideal for research and commercial applications.

A 40B-parameter causal decoder-only model built by TII, trained on 1,000B tokens of RefinedWeb enhanced with curated corpora. One of the top open-source LLMs.

Large Language Model connected with massive APIs — enables LLMs to accurately produce API calls and write programs that interact with the world.

OpenAI's most advanced large multimodal model — accepts image and text inputs, produces text outputs, and exhibits human-level performance on professional and academic benchmarks.

A state-of-the-art language model from Google with improved multilingual, reasoning and coding capabilities — powers many Google AI products. Since succeeded by the Gemini models.

An autoregressive Large Language Model trained on vast amounts of text data using industrial-scale compute — a multilingual, open-access model by BigScience.

An instruction-following LLaMA model fine-tuned by Stanford — trained on 52K instruction-following demonstrations, enabling strong instruction-following behaviour at low cost.

An encoder-decoder model based on the T5 architecture from Google, fine-tuned with the Flan collection — excels at instruction-following and diverse NLP tasks.

An open-source chatbot trained by fine-tuning LLaMA on user-shared conversations collected from ShareGPT — achieves 90%+ of ChatGPT quality in evaluations.

Like ChatGPT but powered by the RWKV (100% RNN) language model — the only RNN that matches transformers in quality and scaling, while being faster and saving VRAM.

Provides a powerful, open-source base to create both specialised and general-purpose chatbots for various applications — by Together AI.

A free-to-use, locally running, privacy-aware chatbot — no GPU or internet required. Run powerful LLMs on your own hardware with complete data privacy.