More Reasoning Isn’t Always Better
Increasing reasoning effort is not the same as choosing a more capable model: it can introduce mistakes and cost more without improving the answer. The article suggests matching effort to the task, starting low for routine work and increasing it when you can verify an improvement.
Written byIdir Ouhab
These are my personal views and do not represent OpenAI.
You see extra high and switch it on, just to be safe. After all, who wants to choose medium and wonder whether they’re leaving some of AI’s potential on the table?
Over the past few months, I’ve started to worry that reasoning effort is becoming the AI equivalent of “just give me the best one.” We turn it up for everything and assume the answer will be better.
The problem is, we might end up paying more for a worse answer.
Why more thinking can mean fewer correct answers
If the model is more capable and I give it more time, surely it will produce a better answer. Put that way, it sounds perfectly reasonable. It’s a machine, after all.
But how many of us have walked out of an exam knowing that another hour staring at the question wasn’t going to help?
OpenAI showed with o1 that giving models more compute for reasoning improved their performance on difficult problems. Here’s the o1 research, if you can be bothered to read it.
Still, choosing a more capable model and increasing its reasoning effort are two different decisions. With the second, you’re using the same model and giving it more room to explore possibilities and reconsider its answers.
And reconsidering an answer doesn’t necessarily mean improving it. The model might introduce a bad assumption or start paying attention to something it should have ignored.
That’s exactly the kind of behavior described in Inverse Scaling in Test-Time Compute. Some models found the straightforward solution, then talked themselves out of it, got tangled up in irrelevant details, and ended up getting the answer wrong.
In one test involving counting two objects surrounded by nonsensical code, Claude Opus 4 went from almost perfect accuracy to around 80% as reasoning increased.
It needed to count to two. It got carried away.
To be fair, OpenAI o3 improved with more effort on that same test. The effect wasn’t universal, and these tests were designed to probe specific failure modes.
What about high and extra high?
The guidance for GPT-5.5 warns that increasing reasoning effort can reduce quality when instructions conflict, stopping criteria are unclear, or tool access is too open-ended. The model can end up running unnecessary searches and spending too long reasoning through the task.
And yes, if you turn reasoning all the way up and get a worse answer, you still get billed for the API call.
We’ve been here before with context windows
This reminds me of the race to have the biggest context window. And I’m talking about less than a year ago.
Being able to fit more information into a model’s context doesn’t mean it will use that information well.
You probably already know this, but Lost in the Middle showed that the models tested could become less accurate when the relevant information was buried in the middle of a long context. Which, friends, is why Codex introduced auto-compaction.
How to choose a reasoning level
Having different models and reasoning levels lets you match the resources to the job. You don’t need to test every possible combination.
Here’s a simple starting point:
| Use case | Model | Reasoning |
|---|---|---|
| Extracting data, classifying information, and summarizing text | GPT-5.6 Luna | Low |
| Rewriting text and writing simple scripts | GPT-5.6 Luna | Low |
| Everyday coding, reports, and spreadsheets | GPT-5.6 Terra | Medium |
| Comparing sources and preparing proposals | GPT-5.6 Sol | Medium |
| Complex debugging and architecture decisions | GPT-5.6 Sol | High |
| In-depth research across multiple sources with conflicting information | GPT-6 Astra | High |
Turn up the reasoning when you can see what it improves. Your bill doesn’t need any help going up.