Every AI model, explained
GPT, Claude, Gemini, Llama, Mistral, DeepSeek, Qwen, Kimi… Understand what each company’s models do, which one to use, how to access them — and what to know about data and privacy in your region.
🇺🇸 United States
The largest AI labs and the most widely used assistants.
OpenAI
OpenAI makes ChatGPT and offers its models to developers through an API. Its line-up spans flagship GPT models with built-in reasoning, smaller low-cost versions, open-weight models, and dedicated image, video and audio models.
8 model families
Anthropic (Claude)
Anthropic builds the Claude family of AI models, known for careful writing, long-document understanding, strong coding and a focus on safety. Models come in three tiers — Opus, Sonnet and Haiku — refreshed in new generations.
5 model families
Google (Gemini)
Google’s AI models are built largely by Google DeepMind, headquartered in London. Gemini powers Google’s assistant and Workspace; Gemma is its open-weight family; Imagen, Veo and Lyria cover images, video and music.
7 model families
Meta (Llama)
Meta releases the Llama family as open-weight models that anyone can download under Meta’s licence, and uses them in the Meta AI assistant across WhatsApp, Instagram and Facebook.
4 model families
xAI (Grok)
xAI builds Grok, an assistant integrated with the X (Twitter) platform and available as a standalone app, with access to real-time posts on X.
3 model families
Microsoft (Copilot & Phi)
Microsoft brings AI to Windows and Microsoft 365 through Copilot, builds its own compact Phi models, and offers many companies’ models to businesses through Azure.
3 model families
🇪🇺 Europe & 🇬🇧 UK
European AI companies — GDPR, the EU AI Act and data sovereignty built in.
Mistral AI
Paris-based Mistral AI is Europe’s best-known AI model company. It offers both commercial and open-weight models, the Le Chat assistant, and deployment options that suit European data-sovereignty requirements.
8 model families
Black Forest Labs (FLUX)
Black Forest Labs, based in Freiburg, Germany, was founded by researchers behind the original Stable Diffusion. Its FLUX models are among the highest-quality image generators, with some versions released as open weights.
4 model families
Stability AI
London-based Stability AI popularised open image generation with Stable Diffusion and continues to release image, audio, video and 3D models, many with open weights.
3 model families
Aleph Alpha
Heidelberg-based Aleph Alpha focuses on “sovereign AI”: platforms and models that European governments and regulated industries can run under their own control.
2 model families
DeepL
Cologne-based DeepL builds specialised language AI for translation and writing, widely used by European businesses for its quality and privacy standards.
3 model families
🇨🇳 China
Fast-moving labs with many powerful open-weight models.
DeepSeek
Hangzhou-based DeepSeek became famous for high-performing, low-cost models released with open weights. Its models power the free DeepSeek chat app and a low-price API.
4 model families
Alibaba (Qwen)
Alibaba Cloud’s Qwen team releases one of the largest families of open-weight models, from tiny models for phones to large cloud models, plus vision, audio, coding and video models.
5 model families
Moonshot AI (Kimi)
Beijing-based Moonshot AI makes Kimi, an assistant known for long-context reading and research, and has released large open-weight models designed for agentic tasks.
3 model families
Zhipu AI (GLM)
Beijing’s Zhipu AI (also known internationally as Z.ai), a spin-off from Tsinghua University research, develops the GLM family, several of which are released with open weights.
2 model families
MiniMax (Hailuo)
Shanghai-based MiniMax builds language models, the popular Hailuo AI video generator, and speech models used in voice apps.
3 model families
ByteDance (Doubao / Seed)
ByteDance, the company behind TikTok and CapCut, develops the Doubao assistant (hugely popular in China) and the Seed family of language, image and video models.
3 model families
🌍 Rest of the world
Enterprise and multilingual models from Canada, the Middle East and beyond.
Cohere
Toronto-based Cohere focuses on enterprise AI — secure, private deployments for business search, retrieval and agents, with strong multilingual support.
3 model families
TII (Falcon)
The Technology Innovation Institute in Abu Dhabi develops the Falcon family of open-weight models, including versions with strong Arabic support.
2 model families