Key takeaways
- Language models are not random number generators; they repeat the numbers that look random to humans, such as 7, 17, 37, 42 and 73.
- If you need a real random number, ask ChatGPT to run code with a random library, or use one yourself.
- Plus adds higher limits and stronger models over the free tier; Go is a lighter paid tier; Pro is for very heavy professional use.
- Buy a plan only after the free tier's limits actually block your work.
- ChatGPT slows down in very long chats because the browser renders everything and the model rereads it all; start a new chat per task.
ChatGPT picks 73, 17, 37 or 42 when you ask for a random number because language models predict the most likely next token, and in human writing those numbers are the ones people choose when they try to sound random. The model is reproducing our bias, not rolling dice. This page also covers ChatGPT Plus: whether you can get it free, what it offers, how it compares with Go and Pro, whether to buy it, and why ChatGPT can be slow on your laptop or get slower during a chat.
Why does ChatGPT pick 73?
Because it is predicting what a person would most likely write, not generating randomness. In human text, "random" picks cluster on numbers that feel random, such as those ending in 7 and odd primes like 37 and 73, and the model learned that pattern.
When you ask "pick a number between 1 and 100", the model assigns a probability to every possible answer. Training on human writing, plus tuning to give confident answers, concentrates that probability on a few favourites. Sampling at the default temperature then picks one of those favourites most of the time. Ask in a fresh chat and you will often see the same few numbers. That is a useful reminder of how these models work: they produce the most plausible text, which is not the same as the most correct or most random.
Why does ChatGPT always pick 17?
For a 1 to 20 range, 17 plays the same role 73 plays for 1 to 100: it is the number people most often name when asked for a random one, so it dominates the model's predictions. Humans avoid round numbers and the edges of a range, and odd primes feel "more random".
If you need real randomness, do not ask the model to imagine it. Ask ChatGPT to run code instead, for example "use Python's random module to pick a number from 1 to 100", which uses a pseudo-random generator rather than the model's preferences. In your own apps, generate random values in code and pass them to the model if needed. The same lesson applies to anything that must be unbiased, such as shuffling a list, sampling test data or choosing a winner: use code, then let the model explain the result.
Where is ChatGPT Plus free?
ChatGPT Plus is a paid plan and is not generally free in any country. OpenAI has run limited promotions, for example offers for students in some markets, and in India it offered ChatGPT Go, a lighter paid plan, free for a period to eligible users.
Also asked as: where chatgpt plus is free
Promotions are announced by OpenAI and claimed inside ChatGPT or through official partners, never through third-party sites. Beware of "free Plus" offers that ask for your login, sell shared accounts or require an app download: they risk your account and your data, and shared accounts break OpenAI's terms. The free tier is also more capable than many people assume, with access to strong models within limits, so try it seriously before paying.
What does ChatGPT Plus offer?
ChatGPT Plus offers higher usage limits than the free tier, access to more capable models and reasoning modes, and more use of features such as deep research, image generation, advanced voice, file analysis, agent mode and Codex.
The practical gain is fewer interruptions: on the free tier you can run out of the best model and fall back to a lighter one, while Plus lets you work with stronger models for longer. Exact limits and the feature list change often, so check OpenAI's pricing page for the current details. Plus is an individual plan; teams use business plans with admin controls and data protections.
Should I buy ChatGPT Plus?
Buy it if the free tier's limits regularly interrupt work that matters to you, such as daily coding, research or writing, and the time saved is worth the cost. If you use ChatGPT a few times a week for quick questions, the free tier is enough.
A simple test: for two weeks, note every time the free tier blocked you or gave a weaker answer that cost you time. If it happens daily, Plus is likely worth it. Also compare the alternatives: Go if you mostly need higher limits on everyday tasks, and the free tiers of Claude, Gemini and Copilot, which you can combine at no cost. For students, the biggest improvement usually comes from better prompts and checking answers, not from a paid plan.
What is the difference between ChatGPT Go and Plus?
ChatGPT Go is a lower-cost paid plan with higher limits than the free tier for everyday features such as messages, image generation and file uploads. Plus adds access to more capable models and reasoning modes, more advanced features and higher limits again.
Go was launched first in India and then expanded to other countries, aimed at people who use ChatGPT daily but do not need the most advanced models or agent features. Plus suits people who need deep research, stronger reasoning, agent mode or Codex regularly. Because OpenAI adjusts what each plan includes, compare them on the official pricing page for your country before choosing.
What is the difference between ChatGPT Plus and Pro?
Pro is the top individual plan for very heavy professional use. It has much higher limits than Plus and access to OpenAI's most capable reasoning modes, with priority access to new features. Plus is the mainstream paid plan.
Choose Pro only if you hit Plus limits constantly on work that earns money, or you need the strongest reasoning for hard technical, scientific or analytical problems every day. For almost every student and most developers, Plus is the sensible ceiling. Pro does not make ChatGPT infallible; it gives more capacity and the strongest models, but you still need to verify answers.
Why is ChatGPT slow?
ChatGPT is slow when OpenAI's servers are under heavy load, when you use a reasoning model that thinks before answering, or when your conversation is very long. Network problems and browser issues add to it.
Also asked as: why chatgpt is slow
Check these first:
- Status: OpenAI's status page shows outages and degraded performance.
- Model: reasoning modes spend seconds to minutes thinking on purpose; use a faster mode for simple tasks.
- Peak hours: responses slow when many users are active, and free-tier users feel it first.
- Chat length: very long chats take longer to process and render.
- Network: VPNs, proxies and weak Wi-Fi delay streaming responses.
If only one chat is slow, start a new one with a short summary of the earlier context.
Why is ChatGPT so slow on my laptop?
Usually because the browser tab is struggling, not the AI. Very long conversations with code blocks and images create a heavy page, and extensions, low memory or an outdated browser make it worse.
Fixes that work:
- Start a new chat; long chats are the most common cause.
- Close other heavy tabs and apps to free memory.
- Disable extensions, especially ad blockers and page modifiers, or try a private window.
- Update the browser, or try the ChatGPT desktop app.
- Clear the site's cache and cookies if the page behaves oddly.
- Turn off a VPN or switch networks to rule out connection issues.
If ChatGPT is fast on your phone but slow on your laptop, the problem is local to the laptop's browser.
Why does ChatGPT slow down during a conversation?
Because every reply processes the whole conversation so far. As a chat grows, the model has more text to read before answering, and the browser has more messages to render, so each response takes longer.
Long chats also get worse, not only slower: early instructions can be lost or weakened when the context is very long, and the model may mix up details from different parts of the conversation. Keep one chat per task. When a chat gets long, ask ChatGPT for a short summary of decisions and facts, copy it into a new chat, and continue there. For ongoing projects, use a Project in ChatGPT with files and instructions, so you do not need one endless chat.
Sources
GenAI engineer and AI product builder with 2+ years shipping production-grade AI systems: RAG pipelines, fine-tuned LLMs, hybrid retrieval and multi-modal apps across vision, speech and OCR, architected end to end from ingestion to deployment. Leads engineering for SCULT INDIA's 14-member team, founded the 500+ member TechVerse Enclave community and has mentored 200+ students. Three hackathon first prizes: Changethon 2025 (IIT Roorkee), Product Genesis at Vividhotsava 2025 (CSJMU) and BYTEBATTLE (MeetKats).
Open to GenAI roles, hackathon judging, mentorship sessions and guest talks at colleges. For campus sessions the inviting institution covers the visit end to end: travel (a First AC train ticket for journeys up to five hours, a flight for anything longer), accommodation, meals, and pickup and drop; the honorarium and workshop fees are negotiable on WhatsApp or on the call, and requests are welcome from institutions that can cover the visit this way. Email pranjulrathour41@gmail.com.
