What is Claude (Anthropic's LLM)?

Claude is Anthropic's family of AI chatbot models, the same category as ChatGPT or Gemini, but built by a company whose whole reason for existing is making AI development safer.

Who makes Claude, and why is it so focused on safety?

Anthropic was founded in 2021 by a group of researchers who left OpenAI specifically because they wanted to put more weight on safety research than they felt the industry was giving it. That origin still shows up in how Claude acts: it tends to be more careful about risky requests and more willing to say "I'm not sure" than some competitors that lean harder into just being agreeable.

Can you download and run Claude yourself?

No. Unlike Meta's Llama, which publishes its model weights so anyone can download and run it locally, Claude is closed. You only get access through Anthropic's own app, its API, or a cloud platform — there's no version you can download and run on your own hardware.

What is Claude best at?

Where Claude tends to stand out is coding and long documents. It's become one of the most trusted models for software engineering work — reading, writing, and debugging across large codebases — and it can hold onto very long conversations or documents without losing track of earlier details.

Which Claude models are available, and how do you use them?

As of mid-2026, the everyday lineup breaks down like this:

• Claude Opus 4.8 — for the hardest tasks.
• Claude Sonnet 5 — the fast, cheap default that handles most work at near-Opus quality.
• Claude Haiku 4.5 — for speed and low cost.
• Claude Fable 5 — a step-up tier for the most demanding reasoning and long-running agent work.

You can use Claude directly at claude.ai or its mobile apps, build on it through the Anthropic API, or reach it through cloud platforms like AWS Bedrock and Google Cloud if your company already runs infrastructure there.