AI Model Review

What is AI Model Review?

AI Model Review is dedicated to providing transparent reviews and information about generative AI models.

We are currently in the midst of an AI explosion with new generative AI models being released frequently, and by a wide range of organizations, from government institutions, large dominant tech companies to individual creators.

For consumers like us, it can be daunting to understand what the different models have to offer.

While there are information sources about model performance and model rankings, we aim to provide a more qualitative perspective, by asking the models to respond to a standardized set of questions. This way we can directly compare how the models respond and get an intuition about the strengths and weaknesses of different models.

What AI Models Are Available to Compare?

There are currently over 75 models available to compare, and we will continue to add new models as they become available. With the goal of including as many models as possible, our comparisons include both models that can only be accessed via a service such as ChatGPT and models that can be downloaded to run locally like Llama. In the model comparison cards we provide links that will take you to the model page where you can download the weights for the locally available models.

How is Data for the Model Comparisons Generated?

We take our standard set of questions and prompt the models 4 times for each question. Each question is prompted with a different temperature setting. We want to see how the temperature setting affects the model's response. You can change the temperature by selecting from the dropdown at the bottom of the model comparison card to see the different responses.