# Open AI Slashed GPT 5.6 Sol Performance and Told No One — retained record

> Retained reddit record from reddit:OpenAI.

Canonical: https://highsignal.app/data/records/a17c56ff78af2009

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:OpenAI)  published 2026-09-30 06:26 ingested 2026-10-01 02:40

[original source ↗](https://www.reddit.com/r/OpenAI/comments/1wtxliq/open_ai_slashed_gpt_56_sol_performance_and_told/)

## retained text



```
It's kind of ridiculous to be honest, on a model that costs so much, that they would change it a couple weeks after its release, making it barely better than Grok (which costs a fraction of the price) on most tasks. 

I've been using this model with Unity since it came out and was amazed with the results. But over the last 3-4 weeks, I was just appalled with its performance. It had slowly but surely become unable to do even the simplest things, and would often get stuck on infinite loops on complex problems that it knew how to solve when it was first released. 

It's absurd. I know Open AI did this on other models as well, but it was less of a problem when those models didn't cost remotely as much as they do now. Open AI is actively degrading the quality of the LLMs it deems "too expensive" without telling anyone so they can increase their margins after they have gotten people used to using said model. 

Luckily I have bunch of other competitors to choose from. I thought I'd just post for awareness in case other developers find themselves in the same situation today. If work seems to be getting harder all of sudden, it's not in your mind. Companies like these can't be trusted to be transparent in any way shape or form.
```



document fetched 2026-10-01 00:31 · canonical https://reddit.com/r/OpenAI/comments/1wtxliq/open_ai_slashed_gpt_56_sol_performance_and_told
