GPT-4o Mini – Cheaper but Almost As Good Version of GPT-4o
This week, OpenAI introduced a lighter version of its most advanced large language model (LLM), GPT-4o . Called GPT-4o mini, it will replace GPT-3.5 Turbo as the default model used in the free version of ChatGPT, and should be a more affordable option for developers compared to GPT-4o, which OpenAI released a few months ago. back. But exactly how different are these two models and which one should you use? Here’s what you need to know.
A smaller model with some limitations, at least for now.
The biggest compromise with the GPT-4o mini is that it doesn’t yet support video and audio processing at startup, although images do work. OpenAI says it plans to integrate these features later, but for now you’ll have to stick with GPT-4o, which is suitable for them.
Otherwise, the model should achieve performance similar to that seen in GPT-4o, at least according to the charts provided by OpenAI. However, since the way tests are conducted can vary from company to company , some developers prefer to take a wait-and-see approach when it comes to measuring performance.
Prices
Since the GPT-4o mini is designed to perform similarly to the GPT-4o, the biggest long-term difference you’ll see between the two models will be the price, which mostly affects developers rather than individual users. GPT-4o currently costs $5 for 1 million input tokens and $15 for 1 million output tokens. On the other hand, the GPT-4o mini will cost 15 cents per 1 million input tokens and 60 cents per 1 million output tokens, making it much more affordable, especially for longer projects. OpenAI states that 1 million output tokens represented in value is roughly equivalent to approximately 2,500 pages in a standard format book. Individual users simply using the GPT app will still have limited free access to standard GPT-4o, but 4o-mini. will replace the 3.5 Turbo once restrictions on its big brother run out .
Same security features
One area where OpenAI has not deviated from GPT-4o and GPT-4o mini is security. The new, smaller model has the same security mitigations that OpenAI has built into GPT-4o, which includes testing by more than 70 external experts to help identify potential risks the model may pose to its users.
In addition, the GPT-4o mini is the first of OpenAI’s models to implement a new “hierarchical” model, which the company says will help improve the model’s ability to reject rapid injections, hacking attempts, and other nefarious uses such as extracting system hints. .
GPT-4o mini is now available on the ChatGPT app, website and API.