When to use it
Use reasoning Off for quick questions. Increase it when you need help working through a plan, analysing details or understanding an attachment.
OpenAI model in April
An OpenAI model with reasoning settings from Off to Max, so you can choose how much time it spends on a question.
You choose the model and its reasoning setting. Its answer also depends on your question and the April information you let it use.
Use reasoning Off for quick questions. Increase it when you need help working through a plan, analysing details or understanding an attachment.
You can select Luna for chat. April also uses it to analyse attachments. If you choose another chat model, that model receives a description of the relevant content rather than the original file.
OpenAI publishes GPT-5.6 Luna pricing of US$0.20 per million input tokens, US$0.02 per million cached input tokens and US$1.20 per million output tokens.
Max takes longer and uses more of your shared April allowance. It can still give a wrong answer.
Try it for everyday questions or help with an image or file. Check important details in its answer.
These provider prices were checked on 9 September 2026 and may change. April's costs also depend on taxes, routing, caching and request length. You are not billed per token. Your model allowance is shown in the app. Source: OpenAI GPT-5.6 Luna documentation.
The model receives your message and the relevant April information you have enabled. Answers can be incomplete, out of date or wrong. They are not medical advice.
Temporary Chat does not use saved memory or previous chats, and it does not appear in your chat history. An encrypted copy of the response may be kept for up to 24 hours to recover interrupted delivery.
OpenAI analyses attachments first. Your selected model receives a description of the relevant content, rather than the original file.
Check important answers before acting on them. Do not rely on a model to assess an emergency, diagnose a condition, make medication decisions or replace professional advice.
Model and provider names are trademarks of their respective owners. Inclusion does not imply endorsement.