How to Use Claude: Projects, File Uploads and Usage Limits

Pick a plan before you hit the five-hour wall

Claude Free costs $0 and includes Sonnet and Haiku, Artifacts and up to 5 projects, with limited usage every 5 hours and weekly. Pro costs $20 a month, or $17 a month when billed annually at $200 up front, and adds Opus, Claude Code, Claude in Chrome and usage credits. Max is $100 a month and gives 5x Pro usage.

Anthropic's help page says the $20 is the US price and local currency may differ. API use through the Claude Console is billed separately from any of these plans, and there are no standard discounts for individual plans.

Your first chat: choose a model and describe the job

Anthropic describes Sonnet 5.5 as its combination of speed and intelligence and Haiku 5.5 as the fastest model. Opus 5.5 is for long-running coding and knowledge work, and is included from Pro upward. The help pages I read did not list sign-up steps, so this guide leaves those out; claude.ai walks you through them.

Write the task, the audience and the format in your first message. "Summarise this contract for a non-lawyer in 5 bullets" works better than "summarise this", because the model has nothing else to go on. If the answer misses, reply in the same chat with what to change instead of starting over.

What you can upload to one chat

A chat accepts up to 20 files at 500 MB each. Images can be up to 8000 x 8000 pixels. Documents can be PDF, DOCX, CSV, TXT, HTML, ODT, RTF, EPUB, JSON or XLSX, and images can be JPEG, PNG, GIF or WebP. XLSX works only when code execution and file creation are switched on in your account.

PDFs have a page rule that is easy to miss. A PDF of 100 pages or fewer gets both text and visual analysis. From 101 to 1000 pages Claude processes text only, and anything over 1000 pages triggers the error "Uploaded file is too large".

A worked example: a 240-page report that you will question all month

Say you have a 240-page PDF report. In one chat it falls in the text-only band (101 to 1000 pages), so charts are not analysed. That is fine for tables and prose, but not for reading a graph.

Instead, create a Project: open Projects on the left or go to claude.ai/projects, click the new-project button, name it "Q3 report" and click Set project instructions. Write something like "Answer in 3 sentences and quote the page number". Then click the plus button in the knowledge area and add the file.

Project files are limited to 30 MB each, so check that your file is under 30 MB before adding it. Every new chat in the project can now see the report and follow the instructions, so you ask a fresh question on day 2 and day 20 without uploading again.

Projects keep context, but chats inside them do not share it

Free accounts can create at most 5 projects. The project name and description are for you; Claude cannot see them, so put anything it needs in the instructions or the knowledge. Everything in the knowledge area is available to every chat in that project.

Chats in a project do not share context with each other unless the information sits in the knowledge base. If a chat reaches a useful decision, copy it into the knowledge as a text note. On paid plans, when knowledge nears the context window, Claude turns on retrieval (RAG) mode automatically to hold more.

How usage limits work, and how to stretch them

Paid plans have a five-hour session limit, which begins with your first message and resets five hours later, and a weekly limit across all models at a fixed time for your account. Check Settings > Usage to see how much is left and when it resets.

Longer chats use more of your limit. Context size depends on the model: the newest models on paid plans reach 1M tokens, and others have 500K or 200K. Usage in claude.ai, Claude Code and Claude Desktop is counted together. Anthropic gives no message counts, so start a new chat when the topic changes rather than guessing.

Common mistakes

Uploading a 600-page book to a chat and expecting it to read the figures: past 100 pages it reads text only. Putting the same 3 paragraphs of background in every chat instead of in project instructions. Assuming the project description is read by Claude. Ignoring Settings > Usage until a message is refused.

Who Claude suits

It suits people with long documents, repeat tasks that fit a Project, and writing or code help. It is less convenient if you need a tool woven into email and documents, where other assistants are built into the apps. If you are staring at an error on your own screen right now, Crade can read it and explain it, which a chat window cannot do on its own.