Chat Picker

AI Assistant Selection: Use-Case Framework and Recommendations

A hands-on test plan for comparing ChatGPT, Claude, and Gemini across writing, coding, research, and data tasks using only what each vendor documents.

Sources checked 3 Oct 2026

There is no single "best" AI assistant, because each vendor documents different plan prices, usage limits, file capacities, and data settings that matter for different jobs. Chat Picker has not tested ChatGPT, Claude, or Gemini, so this page gives you a plan you can run in an afternoon to see which one fits your work, using only what the vendors publish.

Know these limits before you start

Before you test, note what the vendor pages we read document. Prices below are US dollars as read on October 2026, unless the comparison cell says otherwise.

Plan tiers differ sharply. ChatGPT Free includes unlimited text chats with GPT-5.6 Luna and limited uploads, images, voice, and deep research (OpenAI's ChatGPT pricing page). ChatGPT Plus is $20/month (OpenAI's What is ChatGPT Plus? help page). Claude Free covers chat on web, desktop, and mobile with web search, file creation, and code execution; Claude Pro is $20/month billed monthly or $17/month annual (Anthropic's Claude pricing page). Gemini Free needs a Google Account and includes 15 GB storage; Google AI Plus is $4.99/month and Google AI Pro is $19.99/month (Google's Google AI plans page).

Context windows vary. ChatGPT instant models allow 27K tokens on Free, 54K on Go and Plus, and 128K on Pro; reasoning models allow 256K on Go and Plus and 400K on Pro (OpenAI's ChatGPT pricing page). Claude allows up to 1M tokens on every plan, varying by model (Anthropic's Claude pricing page). Gemini's in-app token figure was not verified on the page we read.

File limits shape what you can upload. ChatGPT caps each file at 512MB and each text or document file at 2M tokens; CSV and spreadsheet files are limited to about 50MB, images to 20MB (OpenAI's File Uploads FAQ). Claude allows 500MB per chat file and 30MB per project file (Claude's upload files help). Gemini allows each non-video file up to 100MB and up to 10 files per prompt (Gemini's file upload help).

Data settings differ. ChatGPT lets you opt out of training on your chats on Free, Go, Plus, and Pro (OpenAI's ChatGPT pricing page). Claude offers opt-out on Free, Pro, and Max, and Team is not trained on by default (Anthropic's Claude pricing page). Gemini's training setting was not verified on the page we read.

Usage-policy points matter for high-stakes work. OpenAI's usage policies prohibit tailored legal or medical advice without a licensed professional (OpenAI's usage policies). Anthropic's usage policy classifies legal interpretation as high-risk and requires a qualified professional to review covered advice before dissemination (Anthropic's usage policy). Gemini's guidelines say it should not generate factually inaccurate outputs that could cause real-world harm to health, safety, or finances (Gemini's policy guidelines). Each assistant's help pages also warn that responses can be incorrect or misleading: ChatGPT notes it may sound confident even when wrong (OpenAI's truth FAQ), and Claude describes hallucinations as a known limitation of current models (Claude's incorrect responses help).

The test plan

Run each step in one assistant first, then repeat in the others so you can compare. Keep a simple log with four columns: task, what you prepared, what you saw, and what you recorded. Do not expect the pages to predict results for you.

  1. Writing and content creation. Prepare a 400-word brief about a product you know. Instruct: "Rewrite this brief for a non-expert audience in 200 words, then list three claims a fact-checker should verify." Look at whether sources or statistics are invented. Record how often you had to correct a fabricated detail.

  2. Coding and software development. Prepare a small function that parses a log file. Instruct: "Write a Python function that counts error lines by hour from this text and returns a summary; explain each step." Look at whether the code runs and whether the explanation matches the output. Record whether you could execute the code in the assistant's environment.

  3. Research and fact-checking. Prepare a specific question with a known answer from a site you trust, such as "What storage does the Gemini Free plan include?" Instruct: "Answer with sources you can open." Look at whether citations point to real pages. Record whether the assistant showed a sources button or citations (Gemini's related sources help).

  4. Data analysis and spreadsheets. Prepare a CSV with about 200 rows of monthly sales data. Instruct: "Upload this file, show the top five rows by total revenue, and create a bar chart." Look at whether charts render and whether numbers match your own tally. Record the file-size limit you hit, if any (OpenAI's data analysis help).

  5. Cost and speed trade-offs. Note each plan's price from the pages above and the usage-reset pattern: Claude resets on a rolling five-hour window (Anthropic's Claude pricing page); Gemini refreshes every five hours up to a weekly limit (Google's Gemini limits page). Record how many messages you sent before a limit notice appeared on your plan.

How to read your results

Your log now holds observed behavior, not scores from anyone else. To decide, read each row against the job you actually do most. If your work is long documents, the context-window and file-size rows weigh more than price. If your work is regulated advice, the usage-policy rows decide: no assistant here should replace a licensed professional, and Anthropic requires disclosed AI involvement for consumer-facing use (Anthropic's usage policy).

Do not average your notes into a ranking. Instead, mark each assistant "fits," "partial," or "no" per task, then count fits for the tasks you run weekly. The assistant with the most fits for your real workload is the one to subscribe to first; you can keep a second for the tasks where the first was "partial."

Where the plans differ

The comparison matrix at Claude vs ChatGPT vs Gemini lists the rows that decide most selections. The rows to check first:

  • Free plan availability and what it includes.
  • Main paid plan price and the limits described (exact message counts are not published by OpenAI or Anthropic).
  • Context window in the app.
  • Training on your chats and how to opt out.
  • File upload size and per-project file caps.
  • Heavy-use and team plans if your volume or group size demands them.

None of these rows names a winner. They show documented differences so you can match a plan to the log you just built.

Sources

For current prices and limits, see the dated comparison pages.

Compare assistants