|
MY AI Tools I use four large language model ("LLM") AI services for the most of the research I do for the website: ChatGPT, Perplexity, Grok, Claude and occassionaly I'll check Gemini. Each of these services offer a variety of models and I am constantly changing my use of those models to fit my needs. I also evaluate new AI services as they become available such as the Deep Seek R1 from China. If I find a new service that I end up using I will rebuild this page accordingly. This field is very dynamic and from month to month my preferred services for certain kinds of research change based upon my new experiences with them. |
I use four Large Language Model ("LLM") AI services to accomplish the bulk of the research I do for my website-published projects. Each of the services provides different capabilities in analysis, depth of research and especially response formatting. Each provider offers a variety of options as to which engine model to use, and some providers offer access to competitors' AI engines at the free level.
For my research purposes, I generally make each topical query to all four AI services, with all four open in its own window (across multiple screens.) I copy/paste the responses into a Google Doc keeping the source format (not plain text!). Then I review the responses as they appear in Google Docs and choose which of the responses to use for that particular topic. In some cases I will pick and choose different pieces of responses from different services or rewrite them entirely using the original response as the basis. I pay close attention to whether the URL links I requested are actually hot and whether the formatting will copy into a web page builder correctly. There is usually significant reformatting I have to do manually to conform to the website rendering. (I use Weebly Pro exclusively.)
Although they are constantly changing, here are some guidelines I have identified from my particular uses:
ChatGPT4-o
My original “go-to” engine and still the first one I always use. I also use its integrated Dall-E image producing tool for simple square images, titles, topic markers etc. $20/mo [GPT means “Generative Pre-trained Transformers”]
Perplexity.AI
I use Perplexity PRO for raw research where I'm looking around a certain topic rather than deep diving in towards a specific answer. Perplexity also provides access to a variety of competitor models, including Deep Seek R1 (too technical for my uses.) I normally go directly to the competitor sites for this so I can have all four responses available in windows at the same time. $20/mo.
Grok
I find myself going to Grok more and more, particularly on the strength of its formatting hot link URLs. Since I actively troll on X, I have access to Grok for just $8/mo. I also really enjoy this bad boy's sense of humor.
Claude
While it is supposed to have a lot of the capabilities of the other engines, the thing I like about Claude is how it can compose prose. It also has a rudimentary graphic builder that I occasionally use. I pay $20/mo. for full access to 3.7 Sonnet but you could get by using their free Claude 3.5 Haiku model.
Gemini
Although I included Gemini in this original analysis below, after my first few months of work I determined that Gemini was not worth spending time on for my purposes.
A note on AI graphics/image tools. When I started this project, I used Midjourney quite a bit and paid for that service. It is very powerful and very interesting to use. However, for in1page.net use I discovered that the simple graphics that I needed were not not only available through the integrated Dall-E service with ChatGPT and the rudimentary graphics services within Claude, but those were actually more effective than Midjourney, especially in terms of accurately integrating text into the image.
Below is an analysis made by each of the five LLM services In response to two queries:
1. What are the various models available using [specific service] and what are their strengths and weaknesses and how much do they cost.
2. Compare and list the strengths and weaknesses of the following large language model AI engines ChatGPT, Perplexity, Grok, Claude and Gemini
For my research purposes, I generally make each topical query to all four AI services, with all four open in its own window (across multiple screens.) I copy/paste the responses into a Google Doc keeping the source format (not plain text!). Then I review the responses as they appear in Google Docs and choose which of the responses to use for that particular topic. In some cases I will pick and choose different pieces of responses from different services or rewrite them entirely using the original response as the basis. I pay close attention to whether the URL links I requested are actually hot and whether the formatting will copy into a web page builder correctly. There is usually significant reformatting I have to do manually to conform to the website rendering. (I use Weebly Pro exclusively.)
Although they are constantly changing, here are some guidelines I have identified from my particular uses:
ChatGPT4-o
My original “go-to” engine and still the first one I always use. I also use its integrated Dall-E image producing tool for simple square images, titles, topic markers etc. $20/mo [GPT means “Generative Pre-trained Transformers”]
Perplexity.AI
I use Perplexity PRO for raw research where I'm looking around a certain topic rather than deep diving in towards a specific answer. Perplexity also provides access to a variety of competitor models, including Deep Seek R1 (too technical for my uses.) I normally go directly to the competitor sites for this so I can have all four responses available in windows at the same time. $20/mo.
Grok
I find myself going to Grok more and more, particularly on the strength of its formatting hot link URLs. Since I actively troll on X, I have access to Grok for just $8/mo. I also really enjoy this bad boy's sense of humor.
Claude
While it is supposed to have a lot of the capabilities of the other engines, the thing I like about Claude is how it can compose prose. It also has a rudimentary graphic builder that I occasionally use. I pay $20/mo. for full access to 3.7 Sonnet but you could get by using their free Claude 3.5 Haiku model.
Gemini
Although I included Gemini in this original analysis below, after my first few months of work I determined that Gemini was not worth spending time on for my purposes.
A note on AI graphics/image tools. When I started this project, I used Midjourney quite a bit and paid for that service. It is very powerful and very interesting to use. However, for in1page.net use I discovered that the simple graphics that I needed were not not only available through the integrated Dall-E service with ChatGPT and the rudimentary graphics services within Claude, but those were actually more effective than Midjourney, especially in terms of accurately integrating text into the image.
Below is an analysis made by each of the five LLM services In response to two queries:
1. What are the various models available using [specific service] and what are their strengths and weaknesses and how much do they cost.
2. Compare and list the strengths and weaknesses of the following large language model AI engines ChatGPT, Perplexity, Grok, Claude and Gemini
Here's what the output from each service looks like copied into a Google Doc. I use this display approach so you can see how the response formatting from each service translated into a Google Doc with a simple copy/paste. Pasting into a Weebly web page results in specific formatting that Weebly injects.