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

A Chinese user says GPT-6 Astra xhigh produced a cost-comparison plan that was still unusable after several revisions, while Opus 5.5 read the codex session and produced a usable result in one shot.

A short comment noting that Anthropic has Opus/Sonnet 5.5 and OpenAI has GPT-6 Astra and .1 Sol, and the best part is being able to use all of them.

A reaction post hoping ultrafast gets cheaper and reaches more subscription tiers, and that Astra 6.1 will be impressive especially at 3D and taste/design.

The author used GPT-6 Astra Max Ultrafast to generate a 3D fountain in 2 minutes 27 seconds for $8.75 and invites questions.

A post argues free ChatGPT users get poor results and assume better models won't be much better, so giving them a small taste of GPT-6.1 Sol or GPT-6 Astra would boost subscriptions.

A Chinese group-chat joke pokes fun at OpenAI rolling out Dots, Space, the Decisions API, GPT-6.1 Sol and a $500 plan while nobody knows where GPT-6.1 Astra went.

An Indonesian-language video post describes dots as a 24/7 AI agent with GPT-6 Astra as its brain, its own cloud computer, access to 4,000+ apps, and the ability to act before you ask, while sensitive actions stay in your hands.

On the same prompt, Astra took 35 minutes, 4.68M tokens and $9.78 to make a glossier, stretchier inflatable toy, while Sol 6.1 took 67 minutes, 7.95M tokens and $2.93 for a close result.

A poster jokes about OpenAI's awkward spot: Sonnet 5.5 now outscores Astra, while 6.1 Astra was not released for safety reasons, and also chides GLM 5.3 for running too fast.

The author shows that models like GPT-6 Astra can now easily generate Minecraft mobs, with a video demo.

From October 30, Pro's Codex usage drops from 20x to 10x Plus and GPT-6 Pro messages fall from 200 to 100 per week, while a new Pro 500 tier launches and existing users receive a one-time 62,500 credits.

OpenAI launched dots, built on GPT-6 Astra, which keeps working on its own cloud PC and opens pull requests via Codex, rolling out to Pro and enterprise plans.