Reasoning Effort
Reasoning effort controls how much the AI thinks before it answers.
Think of it like choosing how carefully someone should work:
- Light is like a quick answer.
- Medium is like a normal careful answer.
- High is like asking the AI to slow down and check its work.
- X-High and Max are for the hardest questions.
Not every model has the same effort choices. Writingmate only shows the effort button when the model can use it.
Higher effort can use more of your message allowance. One question can sometimes spend more than one message, especially when the AI thinks for a long time or writes a long answer.
When should I use each effort?
Light
Use Light for simple tasks.
Good for:
- Rewriting a short sentence
- Making a quick list
- Asking a simple fact
- Drafting a short reply
Light is usually faster.
Medium
Use Medium for most work.
Good for:
- Writing emails
- Summarizing documents
- Planning a project
- Comparing a few options
Medium is the default because it is a good balance between speed and quality.
High
Use High when the answer needs more care.
Good for:
- Hard writing tasks
- Coding help
- Math or logic problems
- Important decisions
- Long document analysis
High can take longer, but it may produce a better answer.
X-High or Max
Use X-High or Max for very difficult tasks.
Good for:
- Complex coding problems
- Deep research questions
- Multi-step planning
- Careful review of important work
These settings can be slower and may use more of your message allowance.
For very hard tasks, one AI turn can sometimes count as more than one message.
Why do some models show different choices?
Different AI models work in different ways.
Some models use effort levels like Light, Medium, and High. Other models use a thinking token budget instead. Some models do not let you change effort at all.
Writingmate checks what the selected model supports and only shows choices that work for that model. For API users, the same information is available from GET /api/openai/v1/models and GET /api/openai/v1/models/{id} in the model's reasoning metadata.
Does higher effort always mean better?
No.
Higher effort is useful for hard tasks, but it is not always needed. For simple tasks, Light or Medium is usually enough.
If the AI is answering something easy, choosing Max may just make the answer slower without making it much better.
What should I pick?
If you are not sure, choose Medium.
Use Light when speed matters.
Use High or above when the task is important, complicated, or needs careful checking.