“I don’t use Astra every day”: OpenAI developer lead tells TNS why

OpenAI recently unveiled a slew of new AI tools and models during its DevDay event. Among these was GPT-6.1 Sol, billed as delivering near-Astra performance at a fifth of the cost. While comparable to GPT-6 Astra on benchmark tests, Sol falls short in certain complex and long-running tasks. Users primarily gravitate toward Sol for their day-to-day needs, using it for coding reviews, message drafting, and other ongoing responsibilities.

GPT-6 Astra, on the other hand, is better suited for adversarial testing and safety checks due to its fresh perspective on the work. Ultrafast mode provides real-time capabilities for critical use cases, offering over 300 tokens per second at a higher price point. OpenAI's developer experience lead explained that while the company ships a lot of updates, the goal is to minimize friction for all users, from casual ChatGPT users to advanced engineers and managers.

Guardrails are in place to control costs and ensure safety, with certain advanced features requiring explicit opt-in.