Since OpenAI launched ChatGPT 2.5 years ago, the selection of large language models (LLMs) available to enhance its popular chatbot has continually expanded.

Now, ChatGPT subscribers at the $20 Plus level and higher can choose from 7 distinct AI models, each possessing unique advantages and disadvantages, when engaging with the chatbot.

The models include:

1. GPT-4o

2. o3

3. o4-mini

4. o4-mini-high

5. GPT-4.5 (Research Preview)

6. GPT-4.1

7. GPT-4.1-mini

But how should a user determine which model to use for their specific prompt, query, or task? Considering you can only select one at a time, the decision is crucial.

AI Scaling Hits Its Limits

Power caps, rising token costs, and inference delays are reshaping enterprise AI. Join our exclusive salon to discover how top teams are:

• Turning energy into a strategic advantage

• Architecting efficient inference for real throughput gains

• Unlocking competitive ROI with sustainable AI systems

Secure your spot to stay ahead: https://bit.ly/4mwGngO

Is help on the way?

Assistance seems to be arriving soon from OpenAI, as recent reports from AI influencers on X, including OpenAI's own researcher “Roon (@tszzl on X)” (believed to be team member Tarun Gogineni), indicate a new "router" function. This function will dynamically select the optimal OpenAI model to respond to user input, based on the content of the input.

As Roon posted on the social network X on July 20, 2025, in a now-deleted response to influencer Lisan al Gaib’s remark that they “don’t want a model router I want to be able to select the models I use”:

“You’ll still be able to select. This is a product to make sure that doctors aren’t stuck on 4o-mini”

Similarly, Yuchen Jin, Co-founder & CTO of AI inference cloud provider Hyperbolic Labs, shared in an X post on July 19.

“Heard GPT-5 is imminent, from a little bird.

• It’s not one model, but multiple models. It has a router that switches between reasoning, non-reasoning, and tool-using models.

• That’s why Sam said they’d “fix model naming”: prompts will just auto-route to the right model.

• GPT-6 is in training.

I just hope they’re not delaying it for more safety tests. :)“

A potentially more advanced GPT-5 model would be monumental news if released, yet the router could significantly enhance the experience for the average ChatGPT user.

This development would align with other third-party products like the web-based Token Monster chatbot, which automatically selects and combines responses from multiple LLMs to address user queries.

When asked about the router concept and comments from “Roon,” an OpenAI spokesperson declined to comment or provide further details at this time.

Solving the overabundance of choice problem

It’s important to note that each time OpenAI has introduced a new LLM to the public, it has thoroughly explained through blog posts or release notes what that specific model is suited for and designed to assist with.

For instance, OpenAI’s “o” series reasoning models — o3, o4-mini, o4-mini-high — have excelled in math, science, and coding tests thanks to benchmarking, while non-reasoning models like GPT-4.5 and 4.1 are better suited for creative writing and communication tasks.

Dedicated AI influencers and power users may have a deep understanding of the capabilities and limitations of these various models.

However, regular users who aren’t as involved in the industry or lack the time and resources to test them all on the same input prompts will likely find it challenging to navigate the multitude of options.

This could result in missing out on more intelligent and capable responses from ChatGPT for their specific tasks. In fields such as medicine, as Roon mentioned, the implications could be significant.

Speculating on how an automatic LLM router might alter public perception and adoption of AI is also intriguing.

ChatGPT had already reached 500 million active users by March. If more users were automatically directed to more advanced LLMs to handle their AI queries, the impact on their workloads and the global economy could be substantial, creating a positive “snowball” effect.

As more individuals and businesses experience the benefits of ChatGPT automatically selecting the right AI model for their queries, the overall acceptance and investment in AI are likely to increase. This would further spread AI-powered workflows worldwide.

Currently, however, this process is somewhat hindered by the requirement for ChatGPT users to be aware of their choice of models and have some understanding of their capabilities. It remains a manual process.

Much like being overwhelmed by the options at a supermarket, the typical ChatGPT user faces an overabundance of choice.

Hopefully, an OpenAI router will effortlessly guide users to the most suitable model for their needs, just like a helpful shopkeeper assisting with product selection.