OpenAI Expands GPT-6 Lineup with Cheaper Sol and Luna Models

The Sol and Luna series occupy distinct niches within OpenAI’s architectural hierarchy. Sol remains the specialized engine for complex technical operations such as software development, while Luna focuses on high-volume clerical tasks like document summarization and data extraction. According to internal benchmarks, the new Sol model achieves reliability metrics comparable to the flagship Astra, cutting the error rate of its predecessor by approximately 50 percent.

This release highlights the aggressive pace of the current AI arms race. The update arrived mere minutes after Anthropic launched its own revised Opus 5.5 model, signaling an escalating battle for dominance in the enterprise and developer markets. OpenAI attributes the reduced costs to improved caching techniques, a shift that positions these models as direct competitors to Anthropic’s Fable and Opus offerings. The new versions are currently rolling out to ChatGPT Work, Codex, and API users, with broader availability for free-tier and desktop clients expected throughout the day.