Has anyone successfully accessed or deployed GPT-6 Astra through AI Foundry? I'm seeing it listed in the model leaderboard, but I'm unsure whether it's generally available or still limited to selected Azure accounts and regions.
4 Answers
It has worked for some people in the playground, including web app generation prompts. However, usage is reportedly very expensive, so it’s worth monitoring token consumption carefully. Access may also depend on having a higher Azure subscription tier.
Some users can create a deployment, but requests may still fail with a generic server error. Deployment availability doesn’t seem to guarantee that inference is active yet.
Access still appears to be very limited. There are reports that broader Azure availability may begin soon, but it isn’t reliably accessible for everyone yet.
The model may appear on the leaderboard, but selecting it can result in a “Model not found” message. Being listed doesn’t necessarily mean it’s enabled for your account.
That matches what I suspected—hopefully the wider rollout happens soon.