GPT-6 Astra in Practice: Tests, Use Cases, and Feedback

A post with an image claims GPT-6 Astra is being rolled out to Palantir's star fleet team, suggesting it is entering US government and enterprise settings.

GPT-6 Pro in Chat is now powered by GPT-6 Astra and is available to all Pro, Business and Enterprise users.

The author shows the same image recreated as SVG by two models, Astra on the left and Fable on the right, to illustrate the difference.

The video is not sped up and shows GPT-6 Astra's actual drawing speed; the result doesn't match the reference image, but watching Astra think it got it right is fun.

After trying Astra in ChatGPT Work and Codex, the author finds it cold, more prone to overthinking at higher reasoning levels, quick to take blame for nothing, and disruptive to existing workflows, leaving them unsatisfied as both a relational and coding user.

A developer says they are officially starting system development with GPT-6 Astra from today, with a screenshot showing the actual work.

The poster says Astra is the most expensive model ever benchmarked on this task, at roughly $15 per 30-minute task.

A post claims some Plus and Business users may already have Astra access, asks for screenshots as proof, and predicts a wider rollout today or tomorrow.

A rumor says GPT-6 Astra comes in a base and a Pro version, with Plus users getting 3 to 30 messages per 5 hours and Pro users up to 600, and that OpenAI staff have begun hinting at it.

A developer argues that prominent X users who got unlimited Astra tokens before launch to build games and flight sims are effectively posting paid ads.

The post argues OpenAI has nearly reached strong AGI internally, treats Astra as bordering on weak AGI, and asks whether xAI already has it too.

Asked to settle whether the outer layer counts as pants when wearing two pairs of underwear, GPT-6 Astra Pro and Fable-5.1 Max argue it out in earnest.