ContextMeter – AI Context Window & Token Counter

See how full your AI chat is. Private, local context meter for ChatGPT, Claude and Gemini, with one-click handoff to a new chat.

ContextMeter shows how much of the AI context window your current chat has used, so you know when to start fresh before answers get worse. What you get - A small meter on ChatGPT, Claude and Gemini with the percent used, color-coded green, amber and red. - A details panel: estimated tokens, room left, and which messages take the most space. - Honest labels: every number is an estimate of the visible messages, and it says so. - Handoff: ask the AI for a structured summary, then continue in a fresh chat with one click. You press Send yourself. - Context capsules: save a summary of a chat and reuse it in a new one. - Draft preview: see what an unsent paste would cost before you send it. - Reads your plan and model from the page, so the meter uses the right window. - Works on long chats: Claude removes older messages from the page as you scroll, so ContextMeter remembers what it has seen and offers a full-chat scan. - Settings for your plan or a custom limit, warning levels, theme and position. Keyboard shortcut: Alt+Shift+M. Private by design Everything runs on your device. No account, no analytics, no servers, no network requests. Chat text is never stored; only chat titles, message ids and token counts are kept on your device. Good to know - Counts are estimates: they cover visible message text, not system prompts, emory, tool output, hidden reasoning or attachments. - Context limits vary by plan and change often. Check yours in your account settings. - ContextMeter is not affiliated with OpenAI, Anthropic or Google.