Today, OpenAI introduced GPT-4o Mini which has significantly brought the costs down to 60% cheaper than GPT-3.5 Turbo. Previously, Decosmic was using Claude's Haiku model because it was cheaper than GPT-3.5 Turbo, had a large context window with vision capabilities for the Standard model.
Lineup Refresh
We are replacing the Standard model with GPT-4o Mini. And the Super, from GPT-4o to the new and most powerful Claude 3.5 Sonnet. This significantly brings costs down for our users and provides the same level, and in most cases better performance in most use cases from before.
The Boost model has been removed due to the low utilization and with the unnoticeable cost differences with the Super model, there wasn't a reason to use Boost.
Custom Models
The custom models button no longer show the model name while Standard or Super is selected to avoid any confusion.