Claude vs Microsoft Copilot: the specialist vs the bundle
These two answer different questions. Claude answers "which AI thinks best": the strongest long-form reasoning, the most controllable memory, Claude Code at $20. Copilot answers "which AI is already inside my documents": embedded in Word, Excel, Outlook, and Windows, grounded in your work data, effectively a $10 delta on a Microsoft 365 bundle. If your work lives in Office, Copilot earns its seat; if the thinking itself is the work, Claude is the better instrument. We build NodePad, the third column, for the version of that work that outgrows one linear chat: Claude, GPT, and Gemini on one canvas, each seeing exactly the context you assembled, with none of Copilot's invisible routing or scaffolding.
| dimension | Claude | Microsoft Copilot | NodePad |
|---|---|---|---|
| Price of entry | Free; Pro $20/mo ($17 annual); Max $100 to $200/mo. | Free (15 daily boosts, peak-hour throttling); Copilot Pro is retired; the path is Microsoft 365 Premium at $19.99/mo with Office and 6 TB bundled. | Free on community models; paid plans from $5/mo. |
| Which model answers | The one you picked; Anthropic publishes exact model names. | Undisclosed routing between GPT-5.x and Microsoft's MAI models; the picker has been shrinking, and default modes never say who answered. | The model you picked per message, by exact name. |
| Where it lives | Its own app and site; integrations via MCP connectors. | Inside Word, Excel, PowerPoint, Outlook, Teams, and Windows itself; the strongest distribution in software. | The browser; a dedicated workspace rather than a sidecar. |
| Context and grounding | 200K tokens on consumer plans; Projects hold knowledge bases with retrieval. | Microsoft Graph grounding over your email, meetings, and files (work licenses); plus memory and safety scaffolding you cannot inspect or strip. | Exactly what you assembled on the canvas. Nothing hidden, nothing injected. |
| Memory | The best controls anywhere: view, edit, pause, reset, incognito, on every plan. | Saved memories plus inferred personalization; user-viewable and deletable, admin-controllable at work. | None by design; continuity is the canvas you can see. |
| Branching and organization | Edit and regenerate keep hidden versions behind per-message arrows. | Linear chats in a flat list; folders are an open feature request. Pages and Notebooks organize documents, not conversation paths. | Fork any point onto a visible parallel path; merge chats on one canvas. |
| Usage limits | A conversation budget with no published numbers. | "Higher" and "highest usage for select features"; no published caps, free-tier peak throttling, and AI credits are owner-only on shared plans. | A visible usage meter; add usage or wait for the monthly reset. |
| Stability of the product | Steady consumer app; models improve in place. | Copilot Pro was retired within two years; picker options keep disappearing. Workflows get reshuffled. | A small product with a narrow promise; the canvas is the whole roadmap. |
No credit card to start. Every paid plan reaches every model.