Copilot web chat exposes 2 token quotas

GitHub Copilot Chat on the web now exposes 2 token-budget views—per message and per session—alongside controls to minimize, resume, and revisit conversations. The GitHub changelog says the controls are generally available across every Copilot plan, turning a previously invisible resource into a live meter. For teams already treating AI tokens as a rationed engineering resource, the operator implication is modest but useful: inspect spend at the task level before blaming the model or buying a larger allowance.