OpenAI has announced the replacement of the Operator model based on GPT-4o with a new model powered by o3. The new model emphasizes increasing safety and performance levels with advanced data processing capabilities.
Model Operator Upgrade
OpenAI has replaced the existing Operator model based on GPT-4o with a new one using the o3 architecture. Nevertheless, the API will continue to operate on the previous model. The company stated that the o3 Operator applies a multi-layered safety approach similar to that of 4o.
Enhanced Safety and Performance
The o3 model has been fine-tuned with additional safety data, which should enhance protection against requests for undesirable actions and the search for sensitive data. Compared to the previously existing GPT-4o model, the new version has become less susceptible to vulnerabilities.
Criticism and Cost of the New Model
One critic of the new model is Mike Knoop, co-founder of the Arc Prize Foundation, who pointed out the potential high costs of computations for o3. It is suggested that the costs for solving tasks could be increased by 10 times compared to initial estimates. This may affect the model's accessibility for a wide range of users.
The upgrade of the Operator model to o3 is seen as a move towards improving safety and performance; however, concerns around the cost of its use are being raised. Moving forward, OpenAI will continue to work on enhancements and the provision of new features.
