The companies and their models

Names below are current as of September 2026. This is the fastest moving page on the site. If a name here is not in your app's menu, it has been replaced by something newer, and the shape of the lineup is still right.

Anthropic: Claude

The company behind Claude, founded in 2021 by researchers who left OpenAI with a focus on safety. Claude is known for careful, well written answers, long documents, and being the model many people reach for when writing and thinking matter more than speed.

The lineup: Haiku (small and fast), Sonnet (the everyday default; Sonnet 5 shipped in June 2026), Opus (the big one; Opus 5, July 2026) and Fable (the frontier model for the hardest, longest work; Fable 5.1, September 2026). Anthropic also created MCP, the standard that lets any AI plug into any tool, and gave it away to an open foundation.

Where you use it: claude.ai, the Claude desktop and phone apps, and Claude Code for people who write software. Free plan, then Pro and Max.

OpenAI: ChatGPT

The company that started the wave with ChatGPT in November 2022, and still the most used. Known for a huge feature set: voice, images, search, scheduled tasks, and connections to a long list of apps.

The lineup in 2026 runs on the GPT-5 family: GPT-5.6 in three sizes (Luna, Terra, Sol), GPT-5.5, and the flagship GPT-6 Astra, announced 3 September 2026 for paid plans. Free accounts get the smallest current model.

Where you use it: chatgpt.com and the ChatGPT apps. Free, Go, Plus and Pro plans. Microsoft's Copilot is built on OpenAI's models too, which is how they end up inside Windows and Office.

Google: Gemini

Google's models, built by Google DeepMind, the lab that made AlphaGo. Known for very large context windows, strong image and video understanding, and being built into things you already use: Gmail, Docs, Android, the Google search page.

The lineup: the Gemini 3 generation, with Flash as the fast everyday tier (Google shipped 3.6, 3.7 and 3.8 Flash over the summer of 2026) and Pro as the big one. Gemini 4 has been teased.

Where you use it: gemini.google.com, the Gemini app, and inside Google's own products.

xAI: Grok

Elon Musk's AI company, closely tied to X (formerly Twitter). Known for live access to what is being said on X, a looser tone, and moving fast. The lineup is the Grok 4 series, with Grok 5 in training and repeatedly delayed. Grok added connectors, custom tools and scheduled automations in 2026, so it can run agents like the others.

Where you use it: grok.com, the Grok app, and inside X. Free plan, then SuperGrok.

Meta: Llama

The maker of Facebook and Instagram publishes its models with open weights, meaning anyone can download and run them. Llama 4 is the current generation, with a fully open release in August 2026. You are unlikely to chat with Llama directly, but it powers Meta AI inside WhatsApp, Instagram and Messenger, and a lot of products from other companies.

The open weights crowd

A handful of others matter because they publish models anyone can run, which keeps prices falling for everyone:

• DeepSeek (China). Startled the industry in early 2025 with a model nearly as good as the leaders at a fraction of the cost. Its V4 models in 2026 continued that.
• Mistral (France). Europe's main lab. Mistral Large 3 and Small 4 in 2025 and 2026.
• Alibaba's Qwen (China). A broad, well regarded open family, widely used by developers.

The tools built on top

Not model makers, but names you will hear:

• Cursor. A code editor with AI built in. Where a lot of software is now written. Runs agents and automations, on a paid plan.
• Claude Code. Anthropic's tool for the same job, in a terminal or the Claude desktop app.
• Perplexity. A search engine that answers in sentences with sources.
• Microsoft Copilot. OpenAI models inside Office, Windows and Edge.
• Apple Intelligence. Apple's on-device features, with ChatGPT available for harder questions.

How to choose, honestly

For someone starting out there is no wrong answer among Claude, ChatGPT and Gemini. All three have a free plan, all three are far better than anything from two years ago, and all three can run the agents built here. Pick the one whose app you already have open. If you live in Google's world, Gemini connects to it most easily. If you want the most polished writing and the most careful agent behaviour, that is Claude's reputation. If you want the most features and the biggest ecosystem, that is ChatGPT.

We do not take a side, and the builder on this site writes for all of them.

What to remember

• Five big names: Anthropic (Claude), OpenAI (ChatGPT), Google (Gemini), xAI (Grok), Meta (Llama).
• Every one of them ships a small, a middle and a big model, and a new generation each year.
• Open weights models keep everyone honest on price.
• Pick the app you have. The model inside it is good enough to start.