<?xml version="1.0"?>
<feed xmlns="http://www.w3.org/2005/Atom" xml:lang="en">
	<id>https://tech.uvoo.io/index.php?action=history&amp;feed=atom&amp;title=Context_window</id>
	<title>Context window - Revision history</title>
	<link rel="self" type="application/atom+xml" href="https://tech.uvoo.io/index.php?action=history&amp;feed=atom&amp;title=Context_window"/>
	<link rel="alternate" type="text/html" href="https://tech.uvoo.io/index.php?title=Context_window&amp;action=history"/>
	<updated>2026-05-10T14:54:14Z</updated>
	<subtitle>Revision history for this page on the wiki</subtitle>
	<generator>MediaWiki 1.35.2</generator>
	<entry>
		<id>https://tech.uvoo.io/index.php?title=Context_window&amp;diff=5681&amp;oldid=prev</id>
		<title>Busk: Created page with &quot;Mostly, but not exactly.  Context window means the model’s total working space for one request/conversation turn:  context window = input tokens + output tokens  So if a mod...&quot;</title>
		<link rel="alternate" type="text/html" href="https://tech.uvoo.io/index.php?title=Context_window&amp;diff=5681&amp;oldid=prev"/>
		<updated>2026-05-03T16:04:09Z</updated>

		<summary type="html">&lt;p&gt;Created page with &amp;quot;Mostly, but not exactly.  Context window means the model’s total working space for one request/conversation turn:  context window = input tokens + output tokens  So if a mod...&amp;quot;&lt;/p&gt;
&lt;p&gt;&lt;b&gt;New page&lt;/b&gt;&lt;/p&gt;&lt;div&gt;Mostly, but not exactly.&lt;br /&gt;
&lt;br /&gt;
Context window means the model’s total working space for one request/conversation turn:&lt;br /&gt;
&lt;br /&gt;
context window = input tokens + output tokens&lt;br /&gt;
&lt;br /&gt;
So if a model has a 128K context window, that does not mean you can always send 128K input and still get a big answer. The output also has to fit.&lt;br /&gt;
&lt;br /&gt;
Example:&lt;br /&gt;
&lt;br /&gt;
Model context window: 128K tokens&lt;br /&gt;
Your input:           100K tokens&lt;br /&gt;
Room left:             28K tokens for output&lt;br /&gt;
&lt;br /&gt;
But there may also be a separate max output token cap:&lt;br /&gt;
&lt;br /&gt;
Context window:      128K&lt;br /&gt;
Input tokens:        100K&lt;br /&gt;
Remaining room:       28K&lt;br /&gt;
Max output cap:       16K&lt;br /&gt;
&lt;br /&gt;
Actual max output:    16K&lt;br /&gt;
&lt;br /&gt;
So the terms mean:&lt;br /&gt;
&lt;br /&gt;
Term	Simple meaning&lt;br /&gt;
&lt;br /&gt;
Input tokens	What you send in: prompt, code, files, logs, chat history&lt;br /&gt;
Output tokens	What the model writes back&lt;br /&gt;
Context	The total text the model can “see” while answering&lt;br /&gt;
Context window	The maximum token capacity for input + output&lt;br /&gt;
Token limit	Usually a general term; could mean context limit or output limit&lt;br /&gt;
&lt;br /&gt;
&lt;br /&gt;
So: context/window is not synonymous with input size. It is the total capacity that input and output share.&lt;/div&gt;</summary>
		<author><name>Busk</name></author>
	</entry>
</feed>