You've reached your monthly limit of premium requests
Copilot plans include a monthly quota of “premium requests” (frontier-model chats, agent-mode runs). When it's gone, you're limited to the included base model until the month resets — unless you enable extra billing or run a BYO-key agent in the same editor.
Agent mode is billed per request with model multipliers — a single long agentic session can eat a large share of a month's premium quota.
Copilot's quota can't be redirected — but the same VS Code window can run Cline, Roo Code, or Kilo Code pointed at Standard Compute (OpenAI-compatible, flat price, unlimited). Keep Copilot's completions; move the heavy agent work to an unmetered extension. Guides: /integrations/cline, /integrations/roo-code, /integrations/kilo-code.
No — the quota resets each calendar month and unused requests are lost.
Chat and agent mode fall back to the included base model (or prompt you to buy overage). Code completions keep working.