OpenAI's flagship reasoning and computer-use model — strong at drafting, coding, science, and operating software, and the successor to GPT-5.6 Sol.
Last verified · 2026-10-23 · by Moe Ameen
Published 2026-09-03 · Updated
Short answer
GPT-6 Astra is OpenAI's flagship model, launched September 3, 2026 as the successor to GPT-5.6 Sol. It is built around computer use, meaning it operates browsers, forms, and desktop software, and it scored 99.9% on ARC-AGI-3 against 7.8% for GPT-5.6 Sol. API pricing is $10 per million input tokens and $50 per million output. It writes and reasons but does not generate images or video.
Key takeaways
GPT-6 Astra is OpenAI's flagship model, released on September 3, 2026 as the successor to GPT-5.6 Sol. OpenAI calls it its most intelligent and aligned model and pitches it around a single headline capability rather than benchmark scores: "computer use," the ability to operate a computer the way a person does — clicking through browsers, spreadsheets, forms, and desktop applications instead of only returning text.
President Greg Brockman called it OpenAI's "most intelligent and, also very importantly, our most aligned model yet." Asked about AGI, he said, "For me personally, I do think we're there."
In practice Astra is strong across reasoning, software engineering, cybersecurity, science, and professional work. On reasoning tests it reportedly posts large gains: 99.9% on ARC-AGI-3 versus 7.8% for GPT-5.6 Sol, 97.6% on FrontierMath Tier 4, and 72.6% on the OSWorld 2.0 computer-use benchmark.
Access is phased and safety-gated. OpenAI started with customers in its Daybreak program, then widened access to ChatGPT Plus, Pro, Business, and Enterprise, the OpenAI API, and Amazon Bedrock. Astra met the "Critical" cybersecurity threshold in OpenAI's Preparedness Framework. Standard API pricing is $10 per million input tokens and $50 per million output tokens.
One thing to be clear-eyed about: Astra is a text-reasoning and computer-use model, not a media generator. It writes, reasons, codes, and operates software — it produces no images, no video, no captions, and it publishes to no platform. Treat exact pricing, benchmark figures, and rollout details as a launch-window snapshot and confirm them against OpenAI's own pages.
Think of Astra as the writer's room and the analyst, not the studio. It is excellent at the part of content work that happens before anything is publishable: reasoning through an argument, drafting a tight script, generating five hook variations, synthesizing a messy transcript into a clean outline. What it does not do is turn any of that into a finished post. It records no talking-head video, designs no carousel, renders no quote card, and — for all the "computer use" headlines — it does not log into your accounts and publish on brand. That is the exact handoff Kompozy is built for.
Concretely: take the script, outline, or research Astra produced and drop it into Kompozy as your source. Kompozy rewrites it to your voice through a Persona Brief, then generates the finished formats a reasoning model can't — a captioned Persona Short or a HeyGen avatar video narrating the script, a brand-exact Carousel built from the outline, Quote Graphics pulling the sharpest lines, per-platform Text Posts, a full Blog Article, and an Email Newsletter. Then Autopilot schedules and publishes the whole set across the eight social platforms plus blog and email from one queue, behind a per-post review. The division of labor is clean and repeatable: Astra thinks and drafts, Kompozy produces and distributes — so a stronger model makes your raw material better without changing where the finished content actually gets made and shipped.
A real AI avatar video generated in the Kompozy workspace. Think of it as where a finished script ends up. GPT-6 Astra did not write this one.
Best for:
Not ideal for:
GPT-6 Astra is OpenAI's flagship model, launched September 3, 2026 as the successor to GPT-5.6 Sol. OpenAI positions it around "computer use" — operating browsers, spreadsheets, forms, and desktop apps like a person — and built it for computer use, browsing, software engineering, cybersecurity, science, and professional work. President Greg Brockman said that, for him personally, AGI is already here.
No. Astra is a text-reasoning and computer-use model — it drafts, reasons, writes code, and can operate software. It does not generate video, images, captions, or carousels. To turn its drafts into avatar video, carousels, quote graphics, and other finished media, you pair it with a content engine like Kompozy that produces those formats and publishes them.
Use Astra to draft a script, outline, or research synthesis, then bring that into Kompozy. Kompozy rewrites it in your voice via a Persona Brief, generates Shorts, carousels, quote graphics, text posts, a blog, and a newsletter, and schedules and publishes them across the eight social platforms plus blog and email — so the model handles the thinking and Kompozy handles the making and distribution.
Astra is available through ChatGPT plans (Plus, Pro, Business, Enterprise) and via the API, with standard API pricing of $10 per million input tokens and $50 per million output tokens. OpenAI is rolling it out enterprise-first behind safety gating, so exact pricing and availability are still expanding — confirm current figures on OpenAI's own pages before relying on them.
We checked every figure on this page against the sources listed below on 2026-10-23 and removed or corrected any claim we could not confirm. OpenAI's own announcement page blocked our fetch, so launch details, quotes, benchmarks, and prices were checked against the iClarified and Constellation Research launch reports listed below. Kompozy plan and format details come from kompozy.io and kompozy.io/pricing.
Sources