I did a full-day deep体验 of GPT-5.6-sol.

My attitude roughly went through three stages:
Excitement → Doubt → Understanding and acceptance.

First, let’s talk about something that’s easy to get confused:
The highest reasoning level of GPT-5.6-sol itself is still xhigh. What the UI shows as Max and Ultra mostly refers to the client’s workflow modes—not simply making the model “smarter” by tweaking it alone.

The most complained-about thing about Ultra today is that it really likes automatically dispatching a large number of sub-agents.
Even for a task that isn’t that complex, it may instantly spin up a bunch of agents, consuming a huge amount of Tokens. The whole process looks grand and dramatic, but the actual payoff may not match the cost.

At the beginning, I had doubts too:
Is it really getting work done, or is it just burning Tokens to put on a show?

Later, once I understood how it works, I lowered the reasoning level to High and reran some real tasks—the experience ended up being much better.
It’s faster than GPT-5.5 xhigh, with clearly stronger capability and stability, and Token consumption is nowhere near as outrageous as Ultra.

At least based on my experience today:
GPT-5.6-sol + High is probably the best combination for everyday use right now.
For complex refactors, full-project reviews, and big tasks that require parallel exploration, Ultra can be considered.
For typical development tasks, there’s no need to crank it to the max right away.

This model isn’t unusable, and it’s not necessarily the case that you should blindly set it to the highest tier.
The real issue is that many people still haven’t figured out the differences between Max, Ultra, and the reasoning level, yet they already start treating Ultra as the default mode.

Used correctly, it’s definitely much stronger than 5.5.
Used wrong, you can only watch Tokens leak out.