Summary
OpenAI has added Sol and Luna under GPT-6 Astra, emphasizing lower cost and higher reliability. Sol handles complex tasks such as coding, while Luna targets high-frequency routine work like document summarization and information extraction. The company says that thanks to caching and inference effi…
Key points
- For developers and enterprises using Codex or the API daily, this means more stable output at half the cost, allowing them to reassess which high-frequency workflows are worth fully automating.
- For the first time, frontier models have seen a supply shift of 'same-tier capability at half the price,' which will directly compress the cost structure across the entire AI application layer.
- Developers can bring back batch summarization, extraction, and agent loops previously shelved due to cost, while enterprise procurement teams need to re-benchmark existing model contracts.
Editorial note
This page is Code & Chain's editorial summary of public sources. It may be prepared with AI assistance and published through an automated workflow. Refer to the original sources; this content is not investment, legal, or tax advice.