AI Glossary
A reference dictionary of common terminology in Large Language Model (LLM) tracking, token counts, context sizes, and limit quotas.
Glossary Details
- Coverage: Core Large Language Model concepts
- Intended Audience: Developers, researchers, and AI agents
- Scope: Client-side metrics tracking & handoff terminology
Core Term Definitions
Context Window
The maximum token memory space a model can process in one query. If a conversation exceeds this window, the model cannot read older prompts and replies in the active thread.
Token / Tokenization
Tokens are the basic semantic text segments models read. Words are split into these groups (e.g. "tracking" can be tokenized as "track" and "ing"). 100 words typically represent around 130 to 140 tokens.
Rate Limit
Restrictions applied by providers on session query volumes (e.g., 50 messages per 5 hours). Hitting the rate limit blocks prompt entries until reset times are cleared.
Context Bridge
The specialized handoff feature inside Meter AI that serializes active conversation logs and populates target inputs on ChatGPT, Gemini, or Grok, transitioning users seamlessly.
Sliding Reset Window
Rate tracking calculations based on timestamps of initial prompts. Limits clear dynamically after each message age exceeds the sliding window duration.