Welcome to the MonDive

Today in MonDive, we’re looking at GPT -6 Astra, OpenAI’s latest model for taking on ambitious work across your computer.

We’ll explore its capabilities, benchmarks, and pricing, along with what people are already building and where it could fit into your own work.

Alright, let’s dive in.

100+ ChatGPT Prompts to Revolutionize Your Day

Stop wasting time figuring out what to ask ChatGPT. This free HubSpot guide gives you 100+ proven prompts to speed up writing, research, planning, and daily tasks.

From better outputs to faster workflows, this is a simple shortcut to getting more value from ChatGPT. Instant access. Real use cases. Prompts you’ll actually use.

Why This Matters

You don’t need to know how to code a 3D world to ask Astra to build one.

People are already creating worlds with AI characters that talk and cooperate. In one game experiment, Astra even played what it had built, spotted problems, and went back to make changes.

That last part is what interests me. Getting AI to produce something impressive is one thing. Having it work through the problems that appear afterward is where a much bigger opportunity opens up.

It makes the idea of an AGI era feel a little less distant. You can give Astra an ambitious assignment and, with the right tools connected, let it take on technical work you couldn’t do alone.

You still decide what is worth making and whether the result is good enough. But you no longer have to know every step before you begin.

How Astra Compares With Other Models

On Terminal Bench 4.0, which tests coding and other technical computer tasks, Astra has a clear lead over Sol and Gemini. Claude Fable 5.1 is much closer, scoring roughly 56% against Astra’s 58%. So this is a substantial improvement over OpenAI’s previous model, but a narrower advantage over its closest Claude competitor.

ARC AGI 3 shows the more dramatic gap: Astra reaches 99.9%, compared with 30.2% for Opus 5 and 7.8% for Sol in the published comparison. This test measures how models figure out unfamiliar interactive problems. Astra’s result uses a special adapter setup; its best standard result is 62.7%, so the headline scores should not be treated as an identical setup comparison.

My reading is that Astra makes a strong case for difficult technical work, while Claude remains close competition in coding. These results do not establish a winner for every task.

Access and Pricing

You can access Astra through ChatGPT Work and Codex on eligible paid plans, or pay separately for API usage.

• Plus: $20 per month.

• Pro: $100 or $200 per month, depending on the usage allowance.

• Business: $20 per person per month with annual billing, or $25 with monthly billing. Minimum of two users.

• Enterprise: Custom pricing.

If you use the API, you pay separately for what the model reads and generates:

Astra matches Fable’s base price, while Sol and Gemini cost less per token. The final bill depends on how much work each model needs to finish the task.

What People Are Doing With Astra

1. Create a World With AI Characters

Matt Shumer asked Astra to build a virtual world in Unreal Engine, filled with characters powered by Astra who would cooperate to survive. A day later, he reported hearing them talking to one another.

Astra was involved in both building the world and running its inhabitants. What makes this unusual is that Astra helped create the world, then powered the characters living inside it.

2. Turn a Property Listing Into a 3D Walkthrough

Yunfan Ye gave Astra a Zillow listing and asked it to recreate the house in Blender, then make a promotional walkthrough video using the listing photos as references.

The result gives the property a different kind of presentation without starting with a filmed tour. The first version didn’t get every detail of the house right, but it gave him a walkthrough to review and refine.

3. Build an Anatomy Website You Can Explore

Ashe used Astra to build Human Atlas, an interactive website containing 2,234 selectable anatomical pieces. Visitors can rotate the body, separate structures, and explore its different systems.

The anatomical models come from the existing BodyParts3D dataset. Astra helped turn that material into a searchable visual learning tool, rather than creating the anatomy from scratch.

4. Build and Refine a 3D Game

Anshu used Astra to build a 3D game prototype that runs in a browser. He reported creating the first visual demo in about 45 minutes, then letting Astra work overnight to improve the graphics and add gameplay.

What interests me here is how he directed the work. He gave Astra a visual style to aim for, and it kept refining the scene against that reference. That gives creators a way to develop a game without manually building every detail, though a convincing demo still needs testing before it becomes a polished gaming experience.

5. Create a Website With a 3D Centrepiece

Viktor used Astra to build a website with 3D elements woven into the page design. The result has a distinctive visual style, with depth and detail that go beyond a basic landing page.

What stands out is how the elements work together. The 3D visuals, typography, and layout feel like parts of the same design, showing how much creative direction you can explore with Astra.

6. Generate an Entire Digital Forest

Leon Lin gave Astra about five hours to create a highly detailed forest. According to his post, the scene contains 3,808 trees, 2.5 million grass clumps, and almost 40,000 ferns.

The forest runs in a browser. This was a longer experiment focused on how much detail Astra could build into an environment, with the finished scene available for others to explore.

7. Produce a Science Video From a Topic

Derya asked Astra for a five-minute video explaining T cells and specified Remotion, a tool for creating videos through code.

According to his post, Astra decided what to explain, created visuals using Imagegen, and assembled the video. He handed over a topic rather than a prepared script, leaving Astra to work through both the explanation and its presentation.

Know someone who may be interested?