- ChatGPT Plus vs Free [Video]
- Why Upgrade To ChatGPT Plus?
- Is ChatGPT Plus Faster?
- Dall-E Image Generator Review
- ChatGPT 4 Quality
- Data & Code Analysis
ChatGPT Plus vs Free [Video]
Why Upgrade To ChatGPT Plus?
ChatGPT Plus is the paid version of ChatGPT, it currently costs $20/month + VAT and there are 3 main reasons why the average user would want to upgrade to the premium version:
- Access the latest ChatGPT model – ChatGPT 4 is said to be significantly better than ChatGPT 3.5. It can be used in multiple languages and provide more intelligent output than previous models.
- Access To Dall-E Image Generator – This is similar to mid journey and we will look at it later in the article. Describe an image and it will create it for you.
- Access The Plugin Library – This library of 3rd party plugins allows increasing capabilities such as retreiving information directly from websites when given a URL.
On top of this paid subscribers are contributing to the future development of the product. ChatGPT v5 is rumored to be costing over a billion dollars in computing costs to train the model and your subscription would contribute to pushing the product forwards.
Is ChatGPT Plus Faster?
One of the features I have found most useful and unexpected is that ChatGPT is considerably quicker at “typing”. I was previously using a combination of ChatGPT free version and the OpenAI Playground where you are often hanging around waiting for the prompt to finish before reading or using the output.
OpenAI does not provide specific metrics regarding the exact difference in response times, as it can fluctuate. I would estimate it’s more than twice as fast and feels noticeably quicker which ultimately saves the user time.
To test this I tested a number of different models with the following prompt: “Write a roughly 500 word essay on the future of language models and where we will be in 10 years time”
Dall-E Image Generator Review
I currently have a MidJourneysubscription which costs roughly the same as ChatGPT Plus. I was eager to find out if the latest DALL-E 3 model could compete with MidJourney… and it does, kind of, depending on what you want to use it for.
There’s no doubt that MidJourney still creates the better quality images as we can see from this example:
MidJourney greater capacity to create beautiful images, artistic images and realistic photographic images.
Where Dall-E exceeds is it’s capacity to understand prompts better and iterate on an existing image.
With Midjourney I try and focus on creating short, concise prompts to clearly describe a required image. With DALL-E it works more like ChatGPT where the more information you pass in, the better the output will be.
With Dall-E 3 you only get two options at best to choose from in the output and the second is rarely useful.
If you want to create beautiful artistic images then there is no doubt that MidJourney is the best tool for that job. If you want to create logos, vectors, icons, illustrations, text based images, then Dall-E does come with some advantages.
I’m not ready to cancel my MidJourney subscription just yet because I love using it but probably within a year or two Dall-E will have caught up and ChatGPT plus will fulfill both language and image generation requirements.
I’m going to do further research on this so stay tuned for a midjourney vs dall-e analysis coming soon.
ChatGPT 4 Quality
It’s hard to quantify but there is a noticeable difference in the output quality of the models between GPT-3.5 and GPT-4. I’ve included below snippets from the test prompt “Write a roughly 500 word essay on the future of language models and where we will be in 10 years time” to give some insight into the differences.
In conclusion, the next ten years hold exciting possibilities for language models. They will be more intuitive, specialized, and human-like, revolutionizing various aspects of our lives. From conversational assistants to industry-specific models and educational tools, language models will significantly enhance our interaction with technology. Additionally, they will promote global communication, breaking down language barriers and facilitating cross-cultural collaboration. However, ethical considerations and concerns surrounding bias and privacy will require careful attention as we navigate this transformative future. With responsible development and deployment, language models will undoubtedly play a critical role in shaping a more connected and empowered society.
In conclusion, the coming decade will be a period of significant growth and integration of language models into various aspects of human activity. While the trajectory is promising, it will require careful stewardship to navigate the ethical and societal implications. If managed wisely, the advancements in language models could herald a new era of innovation, creativity, and global collaboration, propelling humanity towards a more interconnected and intelligent future.
GPT-4, in my opinion, sounds more natural and human like. It reads less like it was written by an AI bot and the content it creates is more sensical.
It is a marginal improvement however, the first time I used GPT-3 I was blow away by how capable it was. GPT-4 is a step forwards in terms of understanding, clarity of expression and structure of the outputs.
Like I said it’s hard to quantify…
Data & Code Analysis
One area where ChatGPT Plus really excels is in data analysis. I use spreadsheets and CSV files a lot when researching metrics for digital assets. ChatGPT allows me to do things like reducing the columns, look for anomalies, extract rows based on complex criteria described in every day English.
You could do a lot of these task manually in spreadsheet software or with simple scripts but the ability to just paste in csv data and tell it what you want blows my mind every time.
The models have the super human ability to analyse and process large data sets. The
Note that you have to turn on the experimental feature in settings for “advanced data analysis” to use this model.
I’m a developer and I’m finding myself asking ChatGPT why my code isn’t working more and more often. This is particularly useful for frontend dev work, if you are doing something that has been done before then ChatGPT can probably do 90% of the work for you.
Many of the plugins are built around Python which is popular in data science and AI. If you are building scripts or working with data in any way then I expect it would extremely capable of writing, editing and reviewing Python code bases.
My main line of work is Solidity smart contract development where it is slightly less useful and tends to return generic unhelpful advice when debugging. Again if you are doing something that has been done before like creating a NFT contract then it will be more than capable but I wouldn’t trust it beyond very simple tasks.
One thing it is very useful for is adding Natspec code comments to a contract “make my code less hacky”. This in itself is worth the cost of subscription for me as I can’t imagine doing this manually ever again.
There are probably so many small little use cases like this in everyone’s life where language models can help save valuable time from your work week.
The ChatGPT Plus subscription offers a significant upgrade for users who want the best from the most advanced AI tool I’ve come across.
The speed and efficiency of ChatGPT Plus cannot be understated, providing a more streamlined and responsive experience that can save valuable time for users frequently engaging with the AI tool. This, coupled with the advanced plugin library, dall-e image generator and advanced data analysis model allows users to push the boundaries of what is possible with an AI assistant.
As AI continues to evolve and integrate more deeply into our daily lives and work, staying ahead of the curve with a service like ChatGPT Plus could make all the difference in harnessing the full potential of these transformative technologies.
Is it worth it? In my opinion yes, for almost everyone that uses chatGPT on a daily or weekly basis. For $20/month you are getting a better product with more capabilities and faster execution times.