The free tier is a cap schedule, not a feature list
Roundups of free AI assistants list features. Features are not what runs out. What runs out is allowance: the number of messages before the model downgrades, the hours of the day when free users are throttled, the file size you cannot upload, the research runs you do not get. This article is a cap audit: for each free tier, what the cap is, how it feels in a working week, and the moment the paid tier becomes the honest recommendation. The paid field is graded in Microsoft 365 Copilot alternatives and ranked in best Microsoft Copilot alternatives; this page is only about $0.
One clarification first, because this query arrives in several languages and one of its staged variants was in German: the free assistant market is global and the tiers below are the same tiers a "kostenlos" searcher is looking for. There is no separate German free bench; the vendors price and cap worldwide. This site publishes in English, so the audit is in English, and it covers the intent completely.
| Free tier | The cap that matters | Grounding | Best free use |
|---|---|---|---|
| Copilot Chat (included) | Chat only; no in-app grounding | Your tenant's protection, not your files | Employees already on Microsoft 365 |
| ChatGPT Free | Usage limits; model steps down under load | Uploads within limits | General drafting and questions |
| Claude Free | Usage limits on a rolling window | Uploads and projects within limits | Long-form writing and analysis |
| Gemini Free | Standard usage limits | Google account context; image generation included | Google users; image generation at $0 |
| Perplexity Free | Small daily allowance of Pro searches | Web citations on standard searches | Researched questions with sources |
Copilot Chat: the free tier you may already own
The most under-reported free alternative to Microsoft 365 Copilot is Microsoft's own included tier. Copilot Chat ships with eligible Microsoft 365 plans at no extra cost, with enterprise data protection, in a browser and app surface. Its cap is architectural rather than numerical: it does not ground itself in your documents and does not live inside Word or Outlook the way the $30 add-on does. For an employee whose Copilot use is "ask questions and draft text," the included tier covers it and the add-on is a $360-per-year upgrade to grounding they may never use. For an employee whose use is "summarize my meetings and mail," the cap is total: Chat cannot do it at any message count. Auditing which of those two employees you are is the first item on the checklist below, and it decides whether your free alternative is a downgrade of the same product or a different vendor entirely.
ChatGPT Free and Claude Free: the rolling window problem
Both standalone assistants offer genuinely capable free tiers, and both ration them the same way: a usage allowance on a rolling window, with the model or the priority stepping down when you exhaust it or when demand peaks. ChatGPT Free gives everyone access with usage limits; paid Plus is $20 per month for people who hit the wall daily. Claude Free is the same shape, with Pro at $17 per month billed annually or $20 monthly. In practice the caps feel generous for a few questions a day and tight for a working session: a morning spent iterating on a long document can consume the window by lunch, and the reset arrives on the vendor's schedule, not yours. Claude's free tier earns a specific mention for long-form work, since its writing quality at free tier is the reason many users tolerate the window; ChatGPT's free tier earns its mention for breadth, including voice and image features that arrive free before they arrive elsewhere. Neither free tier grounds itself in your company data; uploads within limits are the bridge, and the bridge has a weight limit.
Gemini Free: the generous outlier
Gemini's free tier is the outlier in this audit: the full Gemini app with standard usage limits, Deep Research access, and image generation at $0 with a Google account, on a 32k context window with 15 GB of storage in the dataset's read. The paid slope above it is also gentle (AI Plus at $4.99 per month in the dataset's consumer read, AI Pro at $19.99). For a user whose assistant work is questions, drafts, and images inside Google's world, the free tier is not a demo; it is the product, with the paid tiers buying headroom. The cap to know is the standard-usage ceiling: sustained heavy use still pushes you toward AI Pro, and the deepest Workspace grounding belongs to the paid Workspace world, not the free consumer app.
Perplexity Free: a ration of research
Perplexity rations differently: basic searches are unlimited and cited, while the deeper Pro searches come in a small daily allowance on the free tier. That shape makes free Perplexity a working tool for the questions it is best at (researched answers with sources you can check) as long as you spend the Pro allowance on the questions that deserve it and let basic search carry the rest. Pro at $17 per month billed annually removes the ration. As a Copilot alternative, free Perplexity is a partial one: it replaces the research slice of assistant use completely and the drafting slice not at all, which is precisely how many teams use it alongside an included Copilot Chat.
When free stops being free
The cap audit has a floor: your time. A professional who loses twenty minutes a day to rolling windows, downgraded models, and re-uploading files is paying roughly a paid tier's price in salary for a $0 tool. The honest thresholds: if you hit a cap more than twice a week, if your work needs grounding in files larger than the upload limits, or if the assistant is inside a client-facing workflow where a throttle becomes a missed deadline, the free tier has stopped being free and the paid field in the flagship guide is the correct next read. Below those thresholds, the free field below is a genuine working bench, and the verdict sorts it by job.
