GPT-6 Astra: OpenAI launches model 120 times cheaper for common tasks

The cost of using AI just dropped dramatically. OpenAI launched new low-cost models designed for specific tasks: GPT-5.6 Luna for quick responses and GPT-6 Astra for more complex processing, with prices starting at $1.20 per million tokens (compared to $15-20 for premium models).

To put it in perspective: if your company processes 1 million words monthly (like reviewing 500 documents of 2,000 words each), with GPT-5.6 Luna you would pay less than $2 monthly. With previous models, that same work cost $20-30.

What are the real use cases? Automatic legal contract review, customer email analysis, support ticket categorization, report summarization, data extraction from documents, code validation. Tasks where you need reliable AI but not maximum creativity.

The cheaper models have a logical limitation: they are less "creative" than premium GPT-6, but more than sufficient for 80% of routine business work. It's like comparing a reliable basic car versus a Ferrari: both get you to your destination, but one costs 20 times less.

The competitive impact is real. Startups and small businesses that couldn't previously automate processes now have economical access to technology that only large companies could afford before. This accelerates AI adoption in sectors like legal, accounting, and customer service.

The trade-off: these economical versions work better with specific tasks. They don't serve "anything", but for what they were designed, they work extremely well.