Learn AI / Compare /Context window vs memory

Context window vs memory

If it has a million-token context window, does it remember me?

The short answer

A context window is how much text the model can look at in one go, and it is emptied when the session ends. Memory is a separate feature some products add, where earlier information is stored outside the model and fed back in. Neither is the model learning about you.

Context window

Large Language Models · Starter

In one sentence: The maximum amount of input and prior output a language model can use at one time while producing the next part of a response.
Why it matters: It affects whether a tool can handle long documents, long chats, and multi-step tasks without dropping important details. That matters for procurement, workflow design, risk assessment, and judging whether an output is complete enough to trust for review.
The everyday version: It is the amount of recent text and instructions the model can look at at one time while answering.
The one thing to remember: The context window is the model's working limit for the current step. Instructions, source material, conversation history and generated output all use the same finite space.
The commonest misconception: A larger context window means the model is smarter. A larger window mainly means it can use more material at once. It may help with long inputs, but it does not automatically improve reasoning quality, factual reliability, or domain judgement.

Read the full page

Long-term memory (in AI systems)

AI Agents and Automation · Intermediate

In one sentence: Stored information an AI system can retrieve across separate sessions to personalise, continue tasks, or reuse past facts and preferences.
Why it matters: A system with long-term memory can feel more useful because it does not start from zero every time. It can continue work across days, reuse stable preferences, and reduce repetitive prompting. But this same feature changes the risk profile. Once information is stored and reused, mistakes can persist, sensitive details can spread too far, and old assumptions can shape new outputs without being noticed. For policy, procurement, and human-rights work, the key questions are governance questions: what may be stored, for how long, with whose knowledge, under what access controls, and with what route for correction, deletion, and human review. If you ignore long-term memory, you may misunderstand what the system is actually using to produce an answer or action.
The everyday version: It is like a saved notebook for an AI system. Some useful details from earlier interactions are written down somewhere outside the chat, so they can be brought back in a later session.
The one thing to remember: Long-term memory is persistent information retrieved across sessions, not a change to the model itself. It improves continuity, but stored items need controls for access, retention, correction and deletion.
The commonest misconception: Long-term memory means the model permanently learned something new. Usually it means the system saved information in an external store and retrieved it later. That can change a future response without changing the model’s underlying trained parameters. Training changes the model itself; memory changes what information is supplied to it at runtime.

Read the full page

Short-term memory (in AI systems)

AI Agents and Automation · Intermediate

In one sentence: Temporary information kept available during a current interaction so an AI system can use recent instructions, results and observations in its next step.
Why it matters: For non-specialists, this term matters because it affects whether an AI system stays consistent during a task. Good short-term memory can help a system remember the latest instruction, use a tool result correctly, and avoid repeating or skipping steps. Weak short-term memory can lead to contradictions, lost constraints, duplicate actions or answers that ignore recent corrections. If you are choosing, supervising or auditing an AI system, this helps you ask practical questions: What does it keep during a session? For how long? Can it be updated, summarised or cleared? Does a human review important steps? Those questions matter for quality, safety and accountability.
The everyday version: It is the temporary notepad an AI system uses during one conversation or task so it can keep track of the latest instructions and results.
The one thing to remember: Short-term memory is temporary working state for the current interaction, often stored outside the model and supplied again on later calls. It supports continuity, but neither guarantees accuracy nor implies immediate deletion.
The commonest misconception: Short-term memory means the model permanently remembers me or my organisation. Usually it refers to temporary information kept for the current task or session. It may be cleared when the session ends unless the system has a separate long-term memory feature or another storage system. Whether anything persists depends on the product design, not the phrase alone.

Read the full page