Disclosure: Some of the links you’ll encounter are affiliate links. If you click and buy something, I’ll get a commission. If you’re reading a review of some precious metals company, please understand that some of the links are affiliate links that help me pay my bills and write about what I love with no extra cost to you. Thank you!
Looking for the latest GPT-4 statistics facts and trends for 2024?
My updated guide has everything you need to know, and more.
Feel free to use the jump links below to go to the section that interests you the most.
Also, all the references and resources I used in crafting my GPT-4 stats guide are listed at the bottom of the page.
Let’s go!
Jump here: Key GPT-4 Statistics Facts and Trends for 2024| GPT-4 Basic Stats and Facts (Launch Date, Training Data, User Demographics, Limitations, Apps Using GPT-4…)|GPT-4 vs GPT-3|GPT-4 Statistics Facts and Trends 2024 FAQ|GPT-4 Statistics Facts and Trends 2024 (Conclusion).
Table of Contents
Key GPT-4 Statistics Facts and Trends for 2024
- Open AI released GPT-4 on 14 March 2024.
- GPT-4 has between 200 billion and 280 billion parameters.
- GPT-4 is a multimodal LLM (large language model) capable of processing both image and text inputs.
- GPT-4 comes as a direct successor to GPT-3.5 which is the technology behind ChatGPT and its current version.
- You can get access to GPT-4 via ChatGPT Plus (costs $20/mo).
- GPT-4 boasts an 82% improvement over ChatGPT (GPT 3.5) in identifying inappropriate requests and implementing enhanced guardrails.
- GPT-4 produces output solely in text format but it can understand both textual and image prompts.
- Some demonstrated applications of GPT-4 include performing tax calculations while citing relevant sources. converting hand-drawn website mockups into functional website code, suggesting recipes based on refrigerator contents…
- GPT-4 exhibits superior performance to ChatGPT (based on GPT 3.5) in most academic and professional exams taken by humans, such as Bar Exams, SAT, GRE etc.
- On the Uniform Bar Exam, GPT-4 scored in the 90th percentile, surpassing GPT-3.5’s score in the 10th percentile.
- ChatGPT plugins have the potential to revolutionize GPT’s capabilities, enabling it to interact with external apps like Code interpreters, Zapier, Wolfram, etc.
- Speculations about GPT-4 having 100 trillion parameters compared to GPT-3’s 175 billion parameters were debunked by Sam Altman, OpenAI’s CEO, though the exact size of GPT-4 remains undisclosed.
GPT-4 Basic Stats and Facts (Launch Date, Training Data, User Demographics, Limitations, Apps Using GPT-4…)
When Did OpenAI Launch GPT-4?
Open AI released GPT-4 on 14 March 2024.
Sam Altman, CEO of OpenAI took to Twitter to announce GPT-4.
What Data Was GPT-4 Trained on?
It is currently unknown what data and dataset GPT-4 was trained on.
According to GPT-4 technical report (linked in the resource section)
“Given both the competitive landscape and the safety implications of large-scale models like GPT-4, this report contains no further details about the architecture (including model size), hardware, training compute, dataset construction, training method, or similar”.
Basically, OpenAI don’t want to share the dataset and number of parameters used to train GPT-4 for fear of competition.
As you can imagine artificial intelligence experts are disappointed by the closed nature of OpenAI’s GPT-4, saying it is slowing down research.
What Was the Process of Training GPT-4?
The process of training GPT-4 was similar to the process of training ChatGPT.
First, the model was trained on hundreds of billions of words (GPT-4 was trained on both data licensed from third-party providers (academic papers, Government docs…) and publicly available data (content on the internet);
and then it was fine-tuned by reinforcement learning with human feedback (RLHF).
Was GPT-4 Really Trained on 100 Trillion Parameters?
GPT-4 wasn’t trained on 100 trillion parameters. Sam Altman debunked that rumour and explained that GPT-4 is not significantly bigger than GPT-3.
Since we know GPT-3 has 175 billion parameters, we can assume GPT-4 has between 200 billion and 280 billion parameters.
In fact, according to George Hotz, founder of the The Tiny Corp:
“GPT-4 has 200B parameters & is a 16-way mixture model w/8 sets of weights”
So the image below is NOT the real comparison between GPT-3 and GPT-4 parameters.
What are Some Limitations of GPT-4?
- GPT-4’s command of languages other than English is severely limited.
- GPT-4 doesn’t update its knowledge in real-time.
- ChatGPT can’t analyze audio or video.
- GPT-4 message limit is 25 messages every 3 hours on ChatGPT Plus.
- GPT-4 can process up to 25,000 words of text.
- GPT-4 hallucinates, exhibits social biases, and gives clearly wrong answers…
- GPT-4 is not good at math making basic arithmetical mistakes.
- ChatGPT is heavily constrained by an extensive set of human-crafted “guardrails” that are put in place to prevent it from being offensive or off the wall.
Which Apps use GPT-4?
Plenty of apps are predicted to start integrating GPT-4 into their systems.
Here are 7 apps already using GPT-4 to enhance existing and offer new services to their customers.
- Duolingo– a language learning app that uses GPT-4 to help students learn languages easier.
- Be My Eyes– app for visually impaired people that uses GPT-4 to take images and analyze them for content and context.
- Khan Academy– they use GPT-4 to enhance learning from course participants.
- Stripe– Stripe has already integrated GPT-3 into their product and now they’re integrating GPT-4. Currently, it is unknown how GPT-4 will be used within the Stripe interface.
- ChatSonic– ChatSonic is a Writesonic App that’s similar to ChatGPT, but better.
- Bing AI Chat– Bing Chat uses GPT-4 as its underlying model. This means using Bing AI Chat is akin to using GPT-4 (read my Microsoft Bing statistics and Bing Chat statistics guides to learn more about Bing AI Chat features).
- Petey– Petey’s integration of GPT-4 enhances this app’s abilities and helpfulness to its users.
Who Can use GPT-4?
Currently, only the ChatGPT+ users have access to GPT-4.
Regular ChatGPT users can sign up for GPT-4 waitlist to get access as soon as possible.
GPT-4 vs GPT-3
GPT-3 was revolutionary, and now GPT-4 is significantly better.
GPT-4:
- gives more precise and relevant answers;
- gives more accurate answers;
- capable of seeing and describing an image in great detail;
- better at math;
- better at text editing, paraphrasing, fixing grammar, spelling and writing style errors;
- better at professional tests and exams.
- better at making jokes (though that doesn’t mean the chatbot can understand humour)
GPT-4 is also significantly better than GPT-3 at avoiding sensitive and inappropriate requests;
According to OpenAI article on GPT-4:
“We’ve decreased the model’s tendency to respond to requests for disallowed content by 82% compared to GPT-3.5, and GPT-4 responds to sensitive requests (e.g., medical advice and self-harm) in accordance with our policies 29% more often”.
| Simulated exams | GPT-3.5estimated percentile | GPT-4 estimated percentile | GPT-4 (no vision) estimated percentile |
|---|---|---|---|
| Graduate Record Examination (GRE) Writing | 4 / 6~54th | 4 / 6~54th | 4 / 6~54th |
| Graduate Record Examination (GRE) Quantitative | 147 / 170~25th | 163 / 170~80th | 157 / 170~62nd |
| Graduate Record Examination (GRE) Verbal | 154 / 170~63rd | 169 / 170~99th | 165 / 170~96th |
| LSAT | 149~40th | 163~88th | 161~83rd |
| SAT Evidence-Based Reading & Writing | 670 / 800~87th | 710 / 800~93rd | 710 / 800~93rd |
| SAT Math | 590 / 800~70th | 700 / 800~89th | 690 / 800~89th |
| Uniform Bar Exam (MBE+MEE+MPT)1 | 213 / 400~10th | 298 / 400~90th | 298 / 400~90th |
GPT-4 Statistics Facts and Trends 2024 FAQ
#1- What Is GPT-4?
GPT stands for Generative Pre-trained Transformers.
GPT-4 is the direct successor to ChatGPT. It’s an LLM (large language model) created by OpenAI (OpenAI statistics) that supports both text and image user inputs and responds with text outputs.
GPT-4 was released on March 14, 2024.
#2- How To Access GPT-4?
GPT-4 is currently available through GPT-4 waitlist and in ChatGPT Plus (paid version of ChatGPT which costs $20/mo).
#3- Does GPT-4 Hallucinate and Invent Facts?
GPT-4 both hallucinates and invents facts. And it’s quite egregious too as it can spit out believable text riddled with errors that only an expert would spot.
Please read this article to see how GPT-4, which is supposed to be a far superior model to GPT-3 and which should be hallucinating way less, still makes up facts, and research paper and even gives nonexistent links as attribution.
#4- Can GPT-4 Really Beat Humans on SATs?
GPT-4 can beat humans on SAT’s.
Please read the section on GPT-3 vs GPT-4 within this post to see how GPT-4 performed on various tests, including SAT’s.
#5- Will GPT-4 Replace Data Scientists?
GPT-4 will not replace data scientists. Instead, it will empower scientists to do better work faster.
But in general, neither GPT-4 nor any other AI system can replace the important role that data scientists play in various industries and sectors..
#6- Is GPT-4 Better at Math Than GPT-3?
GPT-4 is better at math than GPT-3. GPT-3 often made silly math mistakes, and GPT-4 beats it easily.
#7- Will GPT-4 Replace Programmers and Engineers?
GPT-4 will not replace programmers and engineers.
Instead, it will them to do better work faster.
#8- What Is OpenAI Evals?
Evals is a framework for evaluating LLMs (large language models) or systems built using LLMs as components. It also includes an open-source registry of challenging evals.
#9- Is GPT-4 AGI?
GPT-4 is not AGI (artificial general intelligence).
However, a Microsoft research paper claims GPT-4 shows sparks of AGI.
According to them:
“GPT-4 is particularly adept at performing various tasks without specific training and has achieved near-human-level performance in areas like mathematics and coding”
GPT-4 Statistics Facts and Trends 2024 (Conclusion)
My updated guide lists the best and latest GPT-4 statistics facts and trends for 2024.
I hope you enjoyed it because the guide is now over.
During my research, I consulted these resources below:
References:
- 53+ GPT-4 Statistics & Facts: Users, Abilities, Limitations (July 2024)
(https://nerdynav.com/gpt-4-statistics-facts/) - GPT-4 Statistics: 100 Trillion Parameters, Release Date
(https://aigeeked.com/gpt-4-statistics/) - 40+ Mind-Blowing GPT-4 Statistics 2024: [Latest Stats]
(https://blogginglift.com/gpt-4-statistics/) - GPT4: In-depth Guide in 2024
(https://research.aimultiple.com/gpt4/) - OpenAI announces GPT-4, claims it can beat 90% of humans on the SAT
(https://www.cnbc.com/2023/03/14/openai-announces-gpt-4-says-beats-90percent-of-humans-on-sat.html) - GPT-4 Is Coming – What We Know So Far
(https://www.forbes.com/sites/bernardmarr/2023/02/24/gpt-4-is-coming–what-we-know-so-far/) - It’s not just statistics: GPT-4 does reason.
(https://jbconsulting.substack.com/p/its-not-just-statistics-gpt-4-does) - What is GPT-4 and Why Does it Matter?
(https://www.datacamp.com/blog/what-we-know-gpt4) - OpenAI says new model GPT-4 is more creative and less likely to invent facts
(https://www.theguardian.com/technology/2023/mar/14/chat-gpt-4-new-model) - Is GPT-4 a Good Data Analyst?
(https://arxiv.org/abs/2305.15038) - What is GPT-4? The Ongoing Evolution of ChatGPT
(https://summalinguae.com/data/what-is-gpt-4/) - OpenAI research on GPT-4 (https://openai.com/research/gpt-4)
- OpenAI co-founder on company’s past approach to openly sharing research: ‘We were wrong’
(https://www.theverge.com/2023/3/15/23640180/openai-gpt-4-launch-closed-research-ilya-sutskever-interview) - These 7 Apps Are Integrating GPT-4, But How Do They Work?
(https://www.makeuseof.com/apps-integrate-use-gpt4/) - The Future of AI: GPT-3 vs GPT-4: A Comparative Analysis
(https://becominghuman.ai/the-future-of-ai-gpt-3-vs-gpt-4-a-comparative-analysis-5ac525cb7ae1) - Microsoft May Launch GPT-4 For AI-generated Videos Next Week: All You Need To Know
(https://www.news18.com/tech/microsoft-may-launch-gpt-4-for-ai-generated-videos-next-week-all-you-need-to-know-7273573.html) - Kubrickian HALlucinations – Using Chat GPT-4 for Clinical Research Review and Synthesis
(https://www.linkedin.com/pulse/kubrikian-hallucinations-my-deep-dive-using-chat-gpt-4-hymel-md/) - Has GPT-4 Ignited the Fuse of Artificial General Intelligence?
(https://www.datanami.com/2023/03/30/has-gpt-4-ignited-the-fuse-of-artificial-general-intelligence/)
Nikola Roza
Nikola Roza is a blogger behind Nikola Roza- SEO for the Poor and Determined. He writes for bloggers who don't have huge marketing budget but still want to succeed. Nikola is passionate about precious metals IRAs and how to invest in gold and silver for a safer financial future. Learn about Nikola here.