Artificial intelligence is only useful if businesses can actually afford to use it at scale. OpenAI seems to have taken this lesson to heart. Hot on the heels of GPT-6 Astra, its most capable model to date. The company has launched GPT-6 Sol and GPT-6 Luna, two mid-tier models designed to bring near-frontier intelligence to everyday workloads. The headline? API prices have been slashed by 50% compared to their GPT-5.6 predecessors, opening the door for wider, more ambitious AI adoption.
What’s Improved?
GPT-6 Sol and Luna are the latest members of the GPT-6 family. Astra is still the choice for the most complex and demanding tasks. Sol and Luna are designed to be faster, easier to scale and more affordable. Sol’s API pricing has been reduced. Input tokens now cost $2 per million down from $4. Output tokens are now $10 per million down from $20. I think Luna is more budget-friendly. Luna charges $0.10 per million input tokens and $0.50 per million output tokens. Both Sol and Luna are available today, in ChatGPT Work, Codex and the OpenAI API.
What Makes It Different
- Smarter Automation: On AutomationBench GPT-6 Sol does better than Claude Opus 5 costing about 9% of the price per task. GPT-6 Sol can handle multi‑step workflows in sales, marketing, HR and finance.
- Stronger Coding: Sol secures 68.8% on DeepSWE v1.1, which is, near the best rivals and costs about 80% less. Sol also matches Claude Fable 5.1 on FrontierCode benchmarks.
- Sharper Accuracy: Sol makes roughly half as many factual errors as its predecessor, approaching Astra-level reliability.
- Better Caching: Improved prompt caching delivers up to 90% discounts on cached input reads, with a new dashboard to monitor cache performance.
- Refined Communication: A clearer, more direct response style mostly valuable in technical and coding conversations.
How Businesses Use This
For software development companies and enterprises, these releases change the math of AI adoption. Coding agents that once cost thousands per day become economically viable for entire engineering teams. Customer support automation, document processing, and agent-driven business workflows can now run continuously without budget anxiety. Startups and mid-sized businesses, previously priced out of advanced AI, gain access to capabilities that rival models costing several times more.
Conclusion
In this article, GPT‑6 Sol and Luna prove that the future of AI is not about raw power. It is about delivering intelligence affordably and at scale. Whether you are building AI‑powered products or automating operations there has never been a better time to integrate advanced models into your stack.
I think using GPT-6 can boost your project. Contact ScalaCode today. It will help you build GPT-6 AI solutions that fit your business.
