AI Glossary: New Terms for 2026
This is a supplementary glossary covering AI terms that have become common in everyday use but often aren't defined clearly — agentic AI, context window, multimodal, fine-tuning, temperature, and tokens. For foundational terms not covered here, see the main AI Glossary.
Key takeaways
- Agentic AI refers to a model taking multiple steps and actions toward a goal, not just answering a single question.
- Context window is the amount of text a model can 'see' and consider at once when generating a response.
- Multimodal means a model can work with more than just text — images, audio, or video as well.
- Fine-tuning is additional training on a specific, narrower dataset to specialize a model's behavior for a particular task.
- Temperature is a setting that controls how predictable versus varied a model's output is.
Agentic AI
Agentic AI describes a model that takes a series of actions toward a goal rather than just producing one response to one prompt. Instead of only answering a question, an agentic system might search for information, use a tool, check the result, and decide what to do next — multiple steps chained together with limited or no human input between each one. See What Are AI Agents? for a fuller explanation of how this works in practice.
Context window
The context window is how much text a model can consider at once — your current message, any documents or files you've shared, and the conversation history so far. It's measured in tokens, not words. Once a conversation or document exceeds the context window, the model can't see or reference the parts that no longer fit, which is why very long conversations sometimes lose track of details mentioned early on.
Multimodal
A multimodal AI model can handle more than one type of content — for example, understanding text and images together, or generating an image or audio output instead of only text. Early general-purpose AI chat tools were largely text-only; most major tools now handle at least text and images, with some also handling audio or video.
Fine-tuning
Fine-tuning is additional, more focused training applied to an already-trained model, using a smaller and more specific dataset, to adjust its behavior for a particular task or style. A general model might be fine-tuned on a specific company's support transcripts, for example, to make its responses more consistent with that company's tone and typical answers.
Temperature
Temperature is a parameter that controls how predictable or varied a model's responses are. A low temperature makes the model favor its most likely, consistent output — useful for tasks needing precision or repeatability. A higher temperature introduces more variation and creativity, which can be useful for brainstorming but also increases the chance of a less focused or less accurate response. Most everyday chat tools set this for you by default; it's more commonly adjusted when using a model through a developer API.
Token
A token is the basic unit of text a model processes — often a whole word, part of a longer word, or a piece of punctuation. Model context windows, usage limits, and API pricing are typically measured in tokens rather than words or characters, because tokens map more directly to how the model actually processes text internally.
For a system prompt's role in shaping how a model responds, see What Is a System Prompt?, and for terms not covered here, see the main AI Glossary.
Every course and tool mentioned here is included free on AIVX Labs.
Start freeFrequently asked questions
What is agentic AI?
AI that takes a sequence of actions toward a goal — searching for information, using a tool, checking its own work, and moving to the next step — rather than just producing a single response to a single prompt. See What Are AI Agents? for a fuller explanation.
What does context window mean?
The amount of text (measured in tokens) a model can consider at once, including your prompt, any documents you've given it, and its own prior responses in the conversation. A larger context window means it can 'remember' and reference more at once without losing track of earlier parts of the conversation.
What does multimodal mean in AI?
A multimodal model can work with more than one type of input or output — for example, understanding an image you upload in addition to text, or generating an image instead of just text. Earlier AI models were typically text-only; multimodal models handle a mix.
What is a token in AI?
A token is a small chunk of text — often a word or part of a word — that a model processes as its basic unit. Model pricing, context window size, and processing are usually measured in tokens rather than words or characters.
What does temperature mean when using an AI model?
Temperature is a setting that controls how predictable or varied a model's output is. A low temperature produces more consistent, conservative responses; a higher temperature produces more varied, creative — and sometimes less reliable — responses.