Even with 5.2 Thinking (using the Plus subscription), with a complicated prompt, the model doesn't always start the "thinking" process. Why is that? Should i prompt it somehow? Thank you
#Thinking models in ChatGPT
vast idol
neat houndBOT
Hi @vast idol. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:
You’re right that with GPT‑5.2 Thinking on Plus, you won’t always see the visible “thinking” trace, even for prompts that feel complex.
From the docs:
- When you use GPT‑5.2 Auto, it decides whether to use Instant or Thinking based on the task; when it uses reasoning mode, you usually see a slimmed‑down chain‑of‑thought, but “occasionally, a ‘chain of thought’ may not output if the thinking time is low.” [GPT‑5.2 in ChatGPT]
- When you explicitly select GPT‑5.2 with Thinking, there is a thinking‑time toggle (Light/Standard/Extended/Heavy depending on plan). All queries default to Standard, and lower settings favor speed over long visible reasoning. [Thinking time toggle]
So:
- Why doesn’t it always “think” visibly?
- The model may still be doing internal reasoning, but:
- The thinking‑time setting (e.g., Standard/Light) can be too low to show a long chain.
- In Auto mode, it may decide your request doesn’t need full Thinking and stay on Instant. [GPT‑5.2 in ChatGPT]
- The model may still be doing internal reasoning, but:
- Should you prompt it somehow?
- For GPT‑5.2, OpenAI recommends simple, direct prompts; you generally don’t need to say “think step by step” to get reasoning. [Reasoning best practices]
- If you want more visible reasoning in ChatGPT:
- Make sure you’ve manually selected “GPT‑5.2 Thinking” (not Auto).
- In the composer, set thinking time to Extended (or Heavy if you’re on Pro). [Thinking time toggle]