GPT-4o model was recently launched by OpenAI, and now it has announced it has cut the price on the offer significantly. This is true because by embracing the newer version of gpt-4o-2024-08-06, the developers will be charged 50% less on the input tokens $2.50 per million tokens than in gpt-4o-2024-05-13. 33% discount on the output tokens at the rate of $10.00 per million tokens.
GPT-4o model was already fairly cheap to begin with, so this price reduction may relate to increased competition on big LLMs. All of this is supported while delivering about the same level of performance, yet doubling the number of output tokens to 16 as opposed to 4096 in GPT-4o. The current release date for the whole GPT-4o suite, which includes the GPT-4o-f and GPT-4o-q models described in this study, is up to October 2023.
OpenAI releases Structured outputs in the API
In the GitHub release, OpenAI announced the Structured Outputs in the API; the API will guarantee that the model’s generated outputs will fit the JSON Schemas given by developers.
Structured Outputs with response formats are a feature in GPT-4o-mini and gpt-4o-2024-08-06 plus any fine tunes derived from these models. The Availability: This functionality is available on Chat Completion API, Assistants API, and Batch API. Structured Outputs with response formats can also be used with vision inputs.
OpenAI GPT-4o model is cheaper for input and output tokens
As for the inputs, developers reduce their costs by $2,50 for one million input tokens with the assistance of the new gpt-4o-2024-08-06 and save $10,00 per one million output tokens is 33% less than with the help of gpt-4o-2024-05-13. Also, one should understand that JSON Schemas delivered along with Structured Outputs will not apply to Zero Data Retention. Also, there are certain disadvantages when Structured Outputs are used;