Other Large Language Models (LLMs)
Lesson 2 was OpenAI. Here we skim the other names you will see in articles, job posts, and LangChain docs. Still no code — only websites worth bookmarking.
Who else makes LLMs?
ChatGPT is not the whole market. Students often use Claude or Gemini for free. Startups sometimes pick Mistral for cheaper API rates. Meta's Llama models can be downloaded and run offline if you have a decent PC.
| Provider | Chat site | Note |
|---|---|---|
| Anthropic | Claude | Handles long PDFs well |
| Gemini | Works with a Google login | |
| Mistral AI | Le Chat | France-based; popular in EU projects |
| Meta | Llama | Downloadable weights, not just a website chat |
| Cohere | cohere.com | Shows up in search-heavy apps |
An example
I asked "What is the capital of India?" on three sites. The answer matched; only the wording changed.
Three chat tabs, one prompt:
New Delhi.
New Delhi — not Mumbai, which people sometimes guess.
The capital of India is New Delhi.
Pay per message vs run at home
Mistral, Claude, and Gemini all sell cloud access — you send text, they send text back, you get billed by usage. Setup is quick if you have an API key and a card on file.
Llama is different: you can pull the model files and run them with Ollama or Hugging Face. It is slower on an old laptop, but nothing leaves your machine and there is no per-question charge.
This course will use OpenAI first in LangChain — mostly because examples online still assume it. After that works, switching provider is a small config edit, not a rewrite.
What's Next
LangChain — what the library actually does and where it fits with the sites above.