Nabil Luqman

0f9d2dd4f71434084e2b6094de1865e23abb37bb9f28ca51feeabee679ec9be2

Social

artificial intelligence

chatgpt

Open AI

lower cost

Astra 6

powerful

profesional work

OpenAI has officially launched <a href="https://openai.com/index/gpt-6-astra/">GPT-6 Astra</a>, its latest flagship artificial intelligence model and what the company describes as its most capable model broadly deployed to date. Rather than simply generating text in response to prompts, Astra is designed to handle complex tasks from start to finish, including computer use, coding, research, document creation and multi-step professional workflows.
The model is now rolling out gradually, initially reaching selected organisations before becoming more widely available to ChatGPT Plus, Pro, Business and Enterprise users, as well as developers through the OpenAI API, Microsoft Azure and AWS Bedrock.
Astra Is Built To Actually Use A Computer
<iframe src="https://www.youtube.com/embed/1QNsdr-Qx_I" width="560" height="314" allowfullscreen="allowfullscreen"></iframe>
One of the biggest changes with GPT-6 Astra is its improved ability to operate computer interfaces directly.
Instead of simply telling a user how to complete a task, Astra can interact with software and websites to carry out multi-step workflows. OpenAI has demonstrated the model performing tasks such as filling out forms, conducting research, creating presentations and working with specialised technical software.
<img src="https://r2.myc.my/77eb31194b4266dfbc32101b8e10e6f330231d7a9cb729d8f6297fa40fc3ebe7" />
This is part of a broader shift towards more agentic AI, where users can delegate a goal and allow the model to plan, execute and verify multiple steps rather than requiring a new prompt for every action.
OpenAI has also introduced features that allow users to steer Astra while it is working, meaning instructions can be changed mid-task without necessarily restarting the entire process.
Stronger Reasoning And Lower-Cost Task Completion
&nbsp;<img src="https://r2.myc.my/c857cf6a53d2d208c3d4c524cb6a06708539c7870f36926f46e0be7fc88cbc29" />
OpenAI is positioning Astra not only as a smarter chatbot, but as a model capable of handling increasingly difficult technical and professional work.
The company reports a 98% score on FrontierMath Tier 4, 99.9% on ARC-AGI-3 and 100% on ExploitBench, alongside strong results in computer use, software engineering and browsing evaluations.
Astra is also designed to complete complicated tasks using fewer reasoning tokens than previous models, which OpenAI says can help reduce the overall cost of some workloads despite its higher per-token pricing.
For developers, standard API pricing is US$10 per million input tokens and US$50 per million output tokens, while Fast mode costs twice the standard rates. The model also has a 1.05-million-token context window, allowing it to work with very large amounts of information in a single task.
Professional Work Is Where Astra Could Really Matter
<img src="https://r2.myc.my/63b3942a97724552737b30a95e9aad1666634d53e7eef71478acb368f677126f" />
The potential impact of Astra may be most obvious in professional industries where workers regularly spend hours reviewing documents, figures and other repetitive information.
OpenAI highlighted an example involving legal technology company Legora, which used Astra to review 41 documents in a single Agent run within minutes. The system found all four deliberately planted errors in the workflow, including a &pound;500,000 discrepancy, while performance improved by nearly 40% against the previous model on that particular financial-statement task.
Importantly, the example still keeps a human in the loop. The AI carries out the exhaustive comparison, while the professional remains responsible for reviewing the results and making the final judgement.
This distinction could become increasingly important as AI moves beyond generating drafts and begins carrying out entire workflows.
Cybersecurity Raises New Questions
<img src="https://r2.myc.my/ac5f63cd0e408551f5bcad7d8b03b3daa86ea08575305ad89ff5e047f6bd04c8" />
Astra&rsquo;s capabilities are also creating a new set of concerns around cybersecurity.
OpenAI has classified GPT-6 Astra as the first model to reach its &ldquo;Critical&rdquo; cybersecurity capability threshold, meaning that with the necessary tools and access, it can identify previously unknown vulnerabilities and potentially develop ways to exploit them without step-by-step human guidance.
OpenAI says it has introduced stronger protections to prevent harmful cyber activity, including tighter isolation, extensive monitoring and additional alignment evaluations.
The company has also acknowledged a challenge: as models become more capable, understanding exactly what they are doing can become harder. OpenAI chief scientist Jakub Pachocki has warned that the industry may eventually need to slow down deployment or scaling if confidence in safety and monitoring falls too far.
OpenAI Says Astra Could Mark The Start Of The AGI Era
<img src="https://r2.myc.my/9c95f407f50fa526ae8c27d0ec96e4829445f88233ee648cc69022570cf45c09" />
GPT-6 Astra has also reignited the long-running debate around artificial general intelligence (AGI).
OpenAI president Greg Brockman has suggested that future historians could look back on this period as the beginning of the AGI era, pointing to Astra&rsquo;s ability to reason across different domains and carry out economically valuable tasks.
However, that does not mean Astra can autonomously replace humans across the board. Even OpenAI&rsquo;s own professional-work benchmarks show plenty of room for improvement, while real-world applications still require human oversight for important decisions.
The significance of Astra may therefore be less about one sudden moment when AI becomes &ldquo;human-level&rdquo; and more about how quickly AI systems are becoming capable of performing longer, connected chains of work.
Astra Played Through Portal By Itself
<iframe src="https://www.youtube.com/embed/g5u2y0BwRJ0" width="560" height="314" allowfullscreen="allowfullscreen"></iframe>
One of the most striking demonstrations of Astra&rsquo;s capabilities came from an independent experiment involving Valve&rsquo;s Portal, the 3D puzzle game.
AI enthusiast and developer CozyBlaze connected Astra to the game using computer-control tools, allowing the model to receive screenshots, understand the game environment and plan its own movements before executing them.
<blockquote class="twitter-tweet">
And... GPT-6 Astra has autonomously completed Portal! I didn&rsquo;t expect this to happen so soon, but I&rsquo;m glad we've made so much progress here. I was reminded that back in 2016, one of OpenAI&rsquo;s technical goals was to &ldquo;solve a wide variety of games using a single agent.&rdquo; <a href="https://t.co/2nVREdCbMI">pic.twitter.com/2nVREdCbMI</a>
&mdash; cozyblaze (@cozyblazex) <a href="https://x.com/cozyblazex/status/2096383114851533097?ref_src=twsrc%5Etfw">September 5, 2026</a>
<script async="" src="https://platform.x.com/widgets.js" charset="utf-8"></script>
</blockquote>
Astra eventually completed the entire game autonomously over roughly 24 hours, making 3,336 tool calls along the way. The experiment reportedly generated an API bill of around US$571, although CozyBlaze later clarified that the cost was covered through their Codex Pro subscription.
The test required Astra to understand 3D spaces, solve puzzles and continuously interact with the game rather than simply answer questions about it. The game was paused while Astra reasoned, after which the model executed its planned sequence of movements.
The creator behind the experiment stressed that the playthrough should not be treated as a formal AI benchmark. However, it provides an intriguing glimpse at what happens when a general-purpose AI combines reasoning with the ability to actually operate a computer.
And that may be the most important takeaway from Astra: its potential is not simply about producing better answers, but about turning those answers into actions.
Astra Is Already Showing How Different AI Could Become
The Portal experiment, alongside professional workflows, coding and computer-use demonstrations, shows how AI is moving beyond the traditional chatbot model.
Instead of asking an AI one question at a time, users can increasingly give it a broader objective and allow it to figure out the steps needed to complete the job.
That could eventually change how people interact with software itself. Rather than learning how to use complex applications, users could simply describe what they want and have an AI agent operate the tools for them.
There are still clear limitations, safety concerns and reasons for human oversight, but GPT-6 Astra suggests that the gap between AI that can explain a task and AI that can actually perform it is getting smaller.
For now, Astra represents another major step towards AI that does not simply tell you what to do, but actually does it for you.
&nbsp;
*Sources: Visual and Reference Credits to Social Media &amp; various cross-references for context.
FOMO much? Follow our&nbsp;<a style="color: #ff0000;" href="https://www.facebook.com/mycclub/">Facebook</a>&nbsp;and&nbsp;<a style="color: #ff0000;" href="https://www.instagram.com/myc_news/?hl=en">Instagram</a>&nbsp;for more updates.

OpenAI says GPT-6 Astra could mark the beginning of the AGI era. But are we really there yet?

1cc8f807ac47ada93398dcbc39468ea7c51c907358079f69ca0bce2853a44cd7

OpenAI Unveils GPT-6 Astra, Its Most Powerful AI Model Yet

b231707127825f4a15dd17fea55ac4d5a0ca0f639102a0e441db21b9629c225f

AXE MLBB

2af1af6b5e633112ce21f17b1594fa53326dd974471741d13087d637a804a3c9

Cornetto

9621255847abc97937372c8dc13279f044806beb1dd21709fd1553c490afe2dd

BIllie Eilish UMMY

cdb6070aabbec308a543cd4adeeb88164255929fcd5130c1f206502017c1c52f

Coke Fifa

OpenAI Unveils GPT-6 Astra, Its Most Powerful AI Model Yet

OpenAI says GPT-6 Astra could mark the beginning of the AGI era. But are we really there yet?

Latest News

Khairul Aming Responds After Being Labelled A ‘Capitalist’ As Online Debate Shifts To Overconsumption

Proton Unveils New S70 Lite And Prime With Estimated Price Below RM60,000

Urbanscapes Returns To Kuala Lumpur After Seven Years With ‘Made Different’ At Its Core

US Denies Abdul Hadi’s Claim That It Offered PAS Help To Topple Malaysian Government

Haze Worsens Across Malaysia As Indonesia’s Anak Krakatau Erupts And Flights Are Disrupted

Related Articles

Khairul Aming Responds After Being Labelled A ‘Capitalist’ As Online Debate Shifts To Overconsumption

Proton Unveils New S70 Lite And Prime With Estimated Price Below RM60,000

Urbanscapes Returns To Kuala Lumpur After Seven Years With ‘Made Different’ At Its Core

US Denies Abdul Hadi’s Claim That It Offered PAS Help To Topple Malaysian Government

Haze Worsens Across Malaysia As Indonesia’s Anak Krakatau Erupts And Flights Are Disrupted

Related Articles

Khairul Aming Responds After Being Labelled A ‘Capitalist’ As Online Debate Shifts To Overconsumption

Proton Unveils New S70 Lite And Prime With Estimated Price Below RM60,000

Urbanscapes Returns To Kuala Lumpur After Seven Years With ‘Made Different’ At Its Core

US Denies Abdul Hadi’s Claim That It Offered PAS Help To Topple Malaysian Government

Haze Worsens Across Malaysia As Indonesia’s Anak Krakatau Erupts And Flights Are Disrupted