I asked six of the top AI models the same question to compare their response. Some think in databases, some write like pros, and only a few have real-time updates.
Summary: Want the freshest AI data? Copilot and Grok have real-time access—many others work with cutoffs. Claude is great for polished docs, Notion AI turns info into a visual database, and Gemini adds colour-coded highlights. ChatGPT and Meta AI keep it simple. Pick the one that fits how you think!
I was listening to this episode of How I Write and they touched on the difference in knowledge cut-off dates between AI models, and the fact that some have access to real-time data and others don’t — large language models (LLMs) like ChatGPT are trained on a massive dataset of text and code, but this dataset has a specific cutoff date, meaning the model’s knowledge is limited to information available up to that point. In other words, each AI tool has a different point in time where its base knowledge stops, and some compensate for this with access to real-time data while others don’t.
I found this interesting! I’ve only been using ChatGPT so this prompted me to take a “quick look” at which model had the most recent data and, ohhh boy, when I tell you I went on a journey… it was a three hour deep dive into what six top AI tools can do on their free plan.
What started out as a 15 second TikTok idea is now this whole dang article 😝
The Experiment
I asked ChatGPT, Claude, Gemini, Notion AI, Copilot, Meta AI and Grok the same question:
I want to do a TikTok about which AI has the most recent data – can you create a comparison table, including ChatGPT, Claude, Gemini, Notion AI, copilot, meta and grok
For each model, I followed up with requests to:
- Sort by real-time data access capabilities first, then by knowledge cutoff date
- Add pricing information including free tier availability
Let’s look at how each AI performed, with a caveat that it was near impossible to check whether their results were accurate (or at least accurate at the time of their knowledge cut-off), so the whole exercise really validated the recency issue 🙃
This article will probably be out of date by the time you’re reading it, but it’s a fun little tool showdown all the same.
ChatGPT
Knowledge cut-off: October 2023 (GPT-4o)
My thoughts:
ChatGPT’s response formatting was clear and simple, the sources are linked at the bottom, and I liked the at-a-glance visual indication of real time data access, and free tier availability for each tool.
It got its own knowledge cut-off date wrong so that doesn’t bode well for the accuracy of the other data.
Claude
Knowledge cut-off: November 2024 (3.7 Sonnet)
My thoughts:
I liked the way Claude separated the discussion from the table itself, and was impressed by its writing style (having never used it before). The summary dot points at the bottom of the table and in the discussion section were another nice touch. Plus it was helpful to denote the final table as “Latest” with options to copy, download as markdown, or download as PDF.
Once again though, the self-reported knowledge cut-off date was wrong (albeit only by a month).
Google Gemini
Knowledge cut-off: August 2024 (2.0 Flash)
My thoughts:
The Gemini response was very text heavy which made it harder to parse than some of the others, and it didn’t provide the version names or knowledge cut-off dates. I wonder if this was by-design… if it doesn’t provide information unless it’s sure about it? I may be giving too much credit. I liked the fact that it offered notes for my TikTok (part of original prompt), provided disclaimer dotpoints and has the option to export to Sheets, but all in all, it wasn’t my favourite.
Notion AI
Knowledge cut-off: June 2023 (self-reported)
My thoughts:
Notion AI is a bit of a wildcard because it’s built into the notion ecosystem, but I included in my list since I use notion constantly and yet haven’t been using the AI feature. This one surprised me (in a good way!) – I went to ask my starter question as I added a new page, and it inadvertently built out a whole database with fields I hadn’t asked for (like the model version, viral potential and TikTok talking points) and a visual timeline of the cut-off dates for each model, which my brain loved!
It claimed to have a knowledge cut-off of June 2023 but I couldn’t find any official documentation to verify this so, was it the most accurate? We’ll never know*
*Because I’m not about to check when each of the other tools were updated as at June 2023.
Microsoft Copilot
Knowledge cut-off: N/A (continuously updated)
My thoughts:
I liked the way Copilot really stripped things back and got straight to the point, there wasn’t any unnecessary detail. Models with ‘real-time updates’ then didn’t show a knowledge cut-off date, which wasn’t exactly what I asked but I understand how it got there.
I guess it was… fine? I’m not excited to use it again though.
Meta AI
Knowledge cut-off: December 2023
My thoughts:
It was clear, included source information and disclaimers, and left out the fluff, but again seemed to get its own knowledge cut-off date wrong (maybe because it has access to the web for scouting more recent info, and that info might’ve been out of date?!).
Comparing the Approaches
What made this experiment fascinating wasn’t just the data itself, but how each AI approached organizing and presenting information:
| AI Model | Table Organization | User Experience | Standout Feature |
| ChatGPT | Simple, clear | Friendly | Emoji use |
| Claude | Neat, clear | The best of the bunch | Document versioning, output options |
| Google Gemini | Too text-heavy | Not great | Adding a column for TikTok notes, since it was mentioned in my prompt |
| Notion AI | Database-like | Interactive, with multiple views | Timeline visualisation |
| Microsoft Copilot | Simple, boring | Unexciting, with inconsistent choice of notes to include | Brevity |
| Meta AI | Simple, naive | Very basic (no model name) | The list of sources |
Practical Takeaways
Based on this experiment, here’s when I’d use each model:
- For writing: Claude’s writing style, document features and versioning make it an ideal writing partner
- For visualising dates or working within Notion: Notion AI’s database approach was a surprise hit, so I’ll be testing their other AI features (since I use Notion already)
- For quick reference information: ChatGPT provides straightforward answers and is still the one I’m most familiar with
And the rest? I don’t have any use cases for them, for now.
Conclusion
While all models could do the thing I asked, two stood out for different reasons. Notion AI surprised me with its database approach that transformed the comparison into a visual experience. Meanwhile, Claude impressed with its professional writing style, clear document organisation, and practical features like versioning and export options.
Your choice will ultimately depend on how you process information and what you’re using it for, but hopefully this little exploration has given you some clues into the style of each and which one you vibe with.
Have you tried comparing different AI models for the same task? Which one is your go-to?
So far there are six posts in this series, check out the others:
And relatedly: Should I use AI, given its concerning environmental impact?