# accidentally saw “gpt-6 astra” in the normal chat model picker? — retained record

> Retained reddit record from reddit:OpenAI.

Canonical: https://highsignal.app/data/records/9e8fc135296c9d2b

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


source: [reddit](https://highsignal.app/data/reddit) (reddit:OpenAI)  published 2026-09-28 13:32 ingested 2026-09-29 02:39

[original source ↗](https://www.reddit.com/r/OpenAI/comments/1wseqni/accidentally_saw_gpt6_astra_in_the_normal_chat/)

## retained text



```
a couple days ago i randomly got what looked like an experimental model picker in regular chatgpt.

“gpt-6 astra” showed up directly in the chat model list, and it had 3 reasoning modes: light, extended and heavy.

i tried selecting it just to see what would happen, but after sending a message the chat switched me into work mode and the actual model ended up being gpt-6 luna with max reasoning.

it disappeared afterwards and i haven’t been able to reproduce it since.

the weird part is that those light / extended / heavy reasoning labels don’t seem to match the currently documented astra/luna reasoning options, so i’m wondering if i accidentally got some internal/ab-test ui config for a moment.

did anyone else see this recently?
```



document fetched 2026-09-29 00:28 · canonical https://reddit.com/r/OpenAI/comments/1wseqni/accidentally_saw_gpt6_astra_in_the_normal_chat
