LLM
Agents and tools · large language model, LLM model
An LLM (large language model) is a neural network trained on large volumes of text and code that continues text from a prompt; it reads and writes text in tokens. On its own it only replies; to act in your environment it needs a harness, and a model together with a harness is an agent.
The model scores which word comes next. Press ▶
LLM examples: who makes them and which agents run on them
| Models (LLMs) | Agents and tools on top | |
|---|---|---|
| OpenAI | GPT | ChatGPT, Codex |
| Anthropic | Claude: Opus, Sonnet, Haiku | Claude, Claude Code, Claude in Chrome |
| Anysphere | Composer | Cursor (with other vendors’ models too) |
| DeepSeek | the V and R series | its own chat and API; plugged into agents through the API |
| Yandex | Alice AI (formerly YandexGPT) | the assistant and the Alice AI Pro business agent |
| Nous Research | Hermes | Hermes Agent (works with other vendors’ models too) |
Source: Wikipedia: Large language model
Where it appears · 2
- AI agent = LLM + harness · Harness
- Harness engineering in practice · Harness