Copilot Tokenomics (Token Economics)

Track A covers Copilot Tokenomics (token economics) for developers who use GitHub Copilot in VS Code. Every page follows the same shape: Problem → Action → How to verify → Trade-offs.

Check your Copilot billing settings first; the best levers depend on your plan.

Billing model	What is charged	Start with
AI Credits (usage-based)	Input, cached input, and output tokens × their model rates	Measure usage, context, and tools
Legacy Premium Requests	Each prompt/turn × the model multiplier	Billing, models, and budget guards

Agent context is assembled from multiple sources. Source: VS Code Docs — Context in Agents.

Page	Question it answers
How Copilot usage is charged	What am I actually spending?
Measure your usage	Which chats and requests are expensive?
Keep context lean	How do I stop paying for irrelevant history?
Instructions and tools	What is sent on every request, and can it shrink?
Models, planning, delegation	When is a cheaper model or a plan-first flow worth it?
Prompt and output style	How do I shorten prompts and cap answers?
Language and tokenization	Is writing in my language more expensive?
Agent modes and budget guards	Ask vs Plan vs Agent; how do I bound the loop?
MCP audit	Which tool servers should stay enabled?
Checklist	A one-page routine.
cctx: VS Code Token Saver & Accelerator	How do I cut lookup tokens, speed up coding tasks, and see live AIC/USD savings in VS Code?
Resources	Tools and repos worth knowing.
Papers	Research behind these tactics, ranked by authority.

Note

Copilot billing, UI, and settings change quickly. Each page states the date it was verified and links the official source. If something differs in your VS Code, trust the official docs.