If your office runs on Zoho, a dozen stubborn Excel sheets and that one CRM nobody likes, OpenAI's latest update is aimed squarely at you. GPT-6 Astra — the model we've been covering all week — is now generally available inside ChatGPT Work, Codex and the API for paid plans. And the pitch this time isn't "smarter chatbot". It's "AI that works through the software your company already uses, even the stuff with no API at all."
01.What OpenAI actually announced
In an enterprise-focused post this week, OpenAI described Astra as "the world's most intelligent and aligned model," now available in ChatGPT Work, Codex and through the API. The company says it is state-of-the-art on computer use, browsing, professional work, software engineering, cybersecurity and science.
Here's the sentence that should make every small business owner sit up. Most AI systems, OpenAI admits, force businesses to prepare their data, redesign workflows and build custom integrations before delivering any value. Astra is built to skip that. In ChatGPT Work and Codex, it can write code and work through the same applications people use every day — even when those applications don't have an API.
02.A colleague who clicks, types and fills forms
What does "working through applications" mean in practice? Astra uses computer use — it looks at what's on screen, decides the next action, and performs it: clicking buttons, filling forms, copying data between tools. The boring glue work that eats your team's week.
Alongside Astra, OpenAI launched new enterprise plugins in ChatGPT Desktop, powered by the model's browser-use capabilities — Oracle Analytics, Power BI (a Microsoft Fabric service), Navan and Avalara. So your finance team's AI can pull a report out of Power BI roughly the way an intern would, except at 2 a.m. and without complaining.
03.The controls matter more than the capabilities
Handing an AI access to business systems obviously needs guardrails, and OpenAI shipped a batch of them with this rollout. New enterprise admin controls let organisations restrict access to approved websites and desktop applications, manage uploads and downloads, and control browsing history. ChatGPT Work and Codex add confirmation policies that can require human approval before consequential actions, plus automated review of potentially unsafe or unauthorised tool calls.
Notably, enterprise access is off by default — an administrator has to enable it under the organisation's rate card and agreement. That default tells you OpenAI knows exactly how powerful this thing is; we covered Astra's Critical cybersecurity classification in detail earlier, and it's worth reading before you flip any switches.
04.What GPT-6 Astra costs — and how to avoid a bill shock
API pricing starts at $10 per million input tokens and $50 per million output tokens. That's premium territory, so the efficiency claims matter: OpenAI says Astra is trained to complete tasks in fewer tokens with fewer retries — less rework, lower cost per task — and that it now occupies "the majority of the cost-efficiency frontier" on professional work and coding evaluations like Terminal Bench 4.0.
Honestly, this is where most Indian teams will win or lose money. Dollar-denominated token pricing plus agentic workflows that quietly accumulate context across steps is exactly the combination that produces a ₹40,000 monthly API bill nobody budgeted for. If you're planning to build on Astra, the cost-optimization chapter in our GPT-6 Astra — Zero to Hero Guide will pay for itself many times over — it covers prompt caching, context discipline and the settings that keep bills predictable.
05.OpenAI is eating its own cooking
My favourite detail from the announcement: OpenAI rolled Astra out internally weeks before launch. Its developer and marketing teams used Astra and Codex to turn three hours of multicamera footage into a published video that has already crossed 550,000 views in four days. And its engineering team used the model to uncover and fix a memory-allocation bottleneck that was slowing Codex sessions — switching allocators produced 25× lower turn latency with roughly 30% higher peak memory use.
That second story is the real one, I think. An AI that can debug infrastructure problems in its own product is an AI that can debug yours.
06.A quick reality check
Before you rebuild your quarter's roadmap around this: the rollout hasn't been perfectly clean. Reporting around mid-September documented users on OpenAI's own Codex repository still unable to reach Astra, along with missing usage resets and confusing rate-limit changes. Availability also varies — paid ChatGPT tiers get it, Free and Go plans don't, and cloud availability runs through the API, AWS Bedrock and Microsoft's Foundry, each with its own pricing quirks.
So the practical advice is simple: verify it's actually live on your plan before you commit, and start with a small pilot before you give it anything consequential. If you're new to the model altogether, our beginner explainer is a good starting point.
Related reads
- GPT-6 Astra Gets 'Critical' Cybersecurity Rating — the security questions around the same launch.
- Unity's Codex Plugin: GPT-6 Astra Builds Games With You — GPT-6 Astra is now building games inside Unity.
- Sam Altman's 'Big Ship' Tease: GPT-6 Sol Rumors Heat Up — where the GPT-6 line goes next.
Key Takeaways
- GPT-6 Astra is generally available in ChatGPT Work, Codex and the API for paid plans
- It can operate everyday software without needing an API — the "integration tax" of enterprise AI just shrank
- New ChatGPT Desktop plugins: Oracle Analytics, Power BI, Navan and Avalara
- Admin controls, confirmation policies and automated tool-call review ship alongside the model
- API pricing: $10 per million input tokens, $50 per million output tokens — plan your costs deliberately
Frequently Asked Questions
What can GPT-6 Astra do at work?
GPT-6 Astra is now generally available in ChatGPT Work, Codex and the API for paid plans. Its headline trick is that it can operate everyday business software even when there is no API to connect to.
How much does GPT-6 Astra cost for enterprises?
API pricing starts at $10 per million input tokens and $50 per million output tokens. New enterprise plugins bring Oracle Analytics, Power BI, Navan and Avalara into ChatGPT Desktop, and enterprise access stays off by default until an admin switches it on.
What This Means for You
For Indian professionals and small teams, this is the moment enterprise AI stops being a big-company luxury. You don't need a data engineering team or an integration budget — you need someone who knows how to direct the model well: write clear instructions, give it the right level of access, and watch the costs. That's a learnable skill, not a computer science degree. If you want that skill structured for you — from prompting fundamentals through computer use, agents and cost control — the GPT-6 Astra — Zero to Hero Guide is built exactly for this, starting at ₹149.