o1/o1-mini and o3-mini models have been introduced.

Please keep in mind: As these are reasoning models, they are slightly slower and take their time to "think". Temperature configuration does not apply for reasoning models. There is a new property "reasoning effort" to allow model to spend high, medium or low effort, by default we have set this to "low".

Finally there is an additional cost attached, so be extra careful when selecting these models, especially o1. When models "think", they consume additional tokens, which are called "reasoning tokens", and they are priced at the same cost as output token cost of respective models. More information can be found on OpenAI documentation.