Best AIs of 2024 Ranked: ChatGPT vs. Gemini vs. Claude vs. Perplexity
A subjective ranking of the best AIs of 2024... ChatGPT's o1-pro takes the cake.
OpenAI is currently in the AI lead with their new top model (o1-pro)… o1 (non-pro) is still very good and much faster than pro, but has less comprehensive CoT (chain-of-thought).
Google stepped their game up massively and put the world on notice that they are no longer in a phase of bloated hyperwoke stagnation… their new models are excellent and highly competitive (despite retaining wokeness).
Claude is focused mostly on “AI safety” and not really serious about competing, but they still have an elite product in Claude Sonnet (the crew there haven’t gotten it through their thick craniums that people would pay more for an unlimited use version… I guess they prefer communism, socialism, EA, etc.).
Perplexity is still the best AI Search because of: (1) the underlying architecture (optimized for search) and (2) integration with Claude 3.5 Sonnet & GPT4o (this makes it slightly better for many searches than Google Gemini & ChatGPT).
That said, in just the past month or so, ChatGPT’s native search got a lot better… answer retrieval/accuracy were really bad in the past, now they’re mostly very good… for certain searches it outperforms Perplexity or ties.
(Related: OpenAI’s new o3 & o3-mini model)
Best AIs of 2024
Included below are my subjective AI rankings for nearly EOY 2024.
I’ve been using nearly every cutting-edge AI since inception – with heaviest usage of ChatGPT (my favorite)… and I primarily use for info organization and advanced logic/reasoning.
o1-pro is currently the undisputed heavyweight champion, but for non-CoT answers, Google’s Experimental model (completely free) is probably the best.
Google’s mainstream AI “Gemini” is arguably the best value for $ (the free version packs serious heat and the paid version packs even more) along with ChatGPT Plus - take your pick.
Claude is great, but its usage limits are trash and the paid version is a bad value for $.
Perplexity Pro (with Sonnet 3.5 integration) is still the best AI Search, but GPT4o is really solid and has gained a lot of ground.
Disclaimer: This was written December 17, 2024. Rankings could change in the blink-of-an-eye at the rate all of these companies are rolling out products. If you want to know which is the best for you – just try them all.
Best AI for Chain-of-Thought (CoT) Advanced Logic & Reasoning (High IQ)
1. o1-Pro
Currently the best by a wide margin (zero competition other than o1). The only major downside is that it takes a while to get certain answers (*tick tock*).
I’ve had a few answers take >5 minutes… the supercluster neural nets going brrr. The good news is that you can pop up a boatload of windows and have them all running prompts simultaneously.
The bad news is that on the mobile version you cannot leave the app (even if it’s running) or o1-pro will stop generating the answer and say something like network error.
ChatGPT should fix this and make it so that even if you leave the app running in the background, it completes any query regardless.
The other downside for many is that it costs $200/mo.
Unless you need advanced logic/reasoning, it’s 100% not worth the money. Your average random won’t be able to use this model effectively.
In fact, your average random probably won’t notice a difference between ChatGPT4o and o1-pro due to their queries (and will complain, probably even laugh about how bad it is… saw a lot of this on X).
But the reality with o1-pro: it’s akin to having a ~125-130 IQ assistant 24/7.
Pros: Best AI by far (it’s not even close) for advanced logic/reasoning (if you need this), highest IQ, elite voice/video convos
Downsides: Cost ($200/mo.), no PDF uploads/analysis, time (takes a while to answer)
2. o1
A notch just below o1-Pro. Answers aren’t quite as elite. Most won’t notice much of a difference.
I’ve prompted o1 & o1-pro with the same query and they are quite similar in outputs but o1-Pro is just better… for certain types of queries its a lot better.
o1 is a powerhouse though… don’t think that this is somehow a weak model. It scored like ~120 on an IQ test.
Easily the highest IQ rapid-fire (quick response time) CoT chatbot.
Pros: Speed (rapid responses), second-highest IQ, advanced logic, elite voice/video convos
Cons: Cost for unlimited, no PDF analysis
3. DeepSeek
Decent. Some people claim better than o1. I didn’t think it was even as close to o1 preview but maybe close to o1-mini.
I think Gemini Experimental is better overall (same with Claude Sonnet), but we are talking strictly CoT models here… this is CoT and free with like 50 queries per day.
Pretty impressive for a free model, but not on the same playing field as o1 and o1-pro. Like o1-pro and o1, you can track its “thinking” (baked into the output).
Pros: Free with high daily limits
Cons: Chinese (collecting your data), noticeably worse than o1 models for complex problem solving
Best AI for One-Shot Answers (Zero CoT)
It’s important to underscore that these are all extremely close (neck-and-neck) in capabilities… you could pick one at random and I’d probably be satisfied with the output (at least for Gemini Experimental, Claude, Gemini Flash, and ChatGPT).
The only 2 laggards are Grok & Llama… both of which I expect to catch up soon (likely within the next 6-12 months)… there may be a lot of leapfrogging in 2025.
1. Gemini Experimental 1206 (Google/Alphabet)
I had Claude 3.5 Sonnet ahead of Gemini Experimental models for most of 2024, but the 1206 model is arguably better now.
Scores in Chatbot Arena (LM Arena) for Gemini-Exp-1206 reflect my experience… it beats every other model including Claude.
Pros: Free, no usage limits, customized parameters
Cons: Low token limits per query, convo length limits, no internet integration, woke censorship/filters
2. Claude 3.5 Sonnet (Anthropic)
Has been my favorite model of 2024 by far and my most used besides ChatGPT4o. However, it can no longer compete with o1 models and is probably slightly worse at most queries than Gemini-Exp-1206 except coding.
Pros: Elite & different than others for one-shotting
Cons: Usage limits (low), ultra-woke (refuses to answer a lot), people get banned for almost no reason, bad value for $
3. Gemini 2.0 Flash (Google/Alphabet)
Was behind Claude 3.5 Sonnet for a while, but is arguably ahead now. In my experience it’s better for certain things, but I still prefer Claude.
Pros: Free, fast, constantly improving, internet integration, best value for $ (if paid version)
Cons: Interface (I don’t really like it relative to ChatGPT’s), censorship/woke filters
4. ChatGPT4o (OpenaAI)
The OG and most intensely used. Can’t go wrong. I think it’s maybe a little worse than Sonnet and Gemini for certain queries, but for other queries I’ve found it more detailed and accurate.
Pros: Clean interface, does everything, fast, high usage limits, large context windows
Cons: Slightly worse for certain types of queries (e.g. coding vs. Sonnet)
5. Grok (xAI)
The best of all models at rapid and accurate image generation. Image generation capabilities on Grok are underrated (faster and higher accuracy than every other service).
I like using it more than Midjourney now for images. Has access to real-time data from X as well which can be helpful if you want the most updated info.
I’m expecting the next Grok trained on ~100k GPUs to take a huge step up and compete with GPT4o.
Pros: Real-time info. (from X), cites sources from X (easy to check), maximum truth seeking (not woke), takes user feedback (devs active on X)
Cons: Limited capabilities, clearly behind most in quality
6. Llama (Meta)
Llama is always behind. Worth mentioning because they are trying to compete and are open-source.
I think they’ll get a lot better fast because Zuck is highly competitive, but they have their work cut out for them. I’ve used them but always end up thinking mediocre.
Pros: Improved a lot relative to last year, free to use, open-source, can run locally
Cons: Not the best free AI, woke censorship
Best AIs for Internet Search (2024)
1. Perplexity Pro
Have been using now for ~1.5 years. Love the architecture and interface. Was the most elite search for the first half of 2024, but ChatGPT and Gemini are now highly competitive.
I would not recommend paying for Pro to the average person. Even if you have extra cash to throw around, I wouldn’t get Pro unless you like Perplexity and seriously dislike ChatGPT & Gemini’s internet integrations.
Pros: Can pick between Claude & ChatGPT for search integration, architecture is optimized for search, news featured stories (kind of cool)
Cons: Cost ($200/yr. or something – may not be worth it with GPT4o), shopping (marketed a shopping feature but seemed pretty trash), o1 integration produces inaccurate responses (I always click “skip reasoning”)
2. ChatGPT (SearchGPT)
Initially sucked so much I stopped using it (back in early 2024). ChatGPT then forced “auto” search and responses were occasionally really bad.
Sometimes they still are bad, but the improvements have been noticeable (accuracy & actually addressing the query or multiple questions within a single message).
Pros: Improving rapidly, cites sources (generally good & updated), rapid
Cons: Occasionally cites questionable sources & generates outdated info.
3. Gemini Flash
Gemini has search integration and provides rapid and generally accurate results.
My only issues with it are: (1) inaccurate on occasion (with outdated info.) and (2) source citations (sometimes it doesn’t cite anything relevant so I can’t verify accuracy).
Pros: Free, rapid, generally accurate
Cons: Source citations (sometimes doesn’t tell you where it got info – so you aren’t sure whether it’s actually accurate)
What is the best value AI for the money in December 2024?
Gemini or ChatGPT. Google really stepped their game up this year and is balling out… I didn’t think they’d gain ground on OpenAI this rapidly, but they’re doing it.
The only weakness with Gemini is that they haven’t been able to replicate the Chain-of-Thought (CoT) architecture of o1 and o1-pro… but most people don’t need this.
In fact, the free version of Gemini is insanely valuable… basically like a free public service that Google is providing (conversational AI, image generation, video generation, podcast creation, etc.).
There are so many features baked into Google’s Gemini products that people can’t keep up (it’s one after another and seems like a never-ending feature assault)… whoever is in charge at Google AI is crushing.
OpenAI’s ChatGPT Plus at just $20 remains excellent and you get some o1 plus usage (limited), higher usage of o1-mini (really good model), and 4o with internet search (the internet search has improved a lot and sources are generally good).
Which AI would I pick for most people?
If forced to pick a paid AI: ChatGPT Plus ($20/mo).
I like ChatGPT’s interface the most and find it intuitive and easy-to-use (especially with the drag/drop for PDF analysis vs. uploading to Drive first on Google). You have CoT (o1 & o1-mini) plus Sora plus Search etc.
And it’s not like you can’t use Gemini while paying for ChatGPT… Gemini is still free (both the regular version and experimental version) – so you’re not missing much of anything unless you need a much longer context window from Google and/or slightly better reasoning IDK.
I like the experimental Gemini’s safety settings that can be tweaked to your personal preference, but they still won’t eliminate basic woke filters/censorship (I grilled it with some scientifically-validated but controversial queries and it refused to answer.)
What about best free AI? ChatGPT. But Gemini, Claude, DeepSeek (CoT), Grok (images & updated info), Perplexity (search) are all worth trying.
Which AIs do I enjoy using most?
Regardless of which AIs are the “best” - it’s important to determine which you enjoy using most… some people may love Gemini, others may like ChatGPT, others may have no preference.
There are many people who love Claude for coding, etc.
My picks:
Advanced logic: ChatGPT o1-pro & o1
One-shotting: Claude
Internet Search: Perplexity Pro + Claude integration
Which AIs do you like most in 2024?
If you want to mention what you use these AIs for and which you think is the best – feel free to chime in.
My rankings for Dec 17: GPT Pro, GPT or Gemini, Claude – then the others.
Grok for image gen and updated X info. Perplexity Pro and GPT4o for search.