OpenAI's GPT-6 Sol and Luna Arrive at Half the Price of Their Predecessors

OpenAI has rounded out its GPT-6 generation with updated versions of Sol and Luna, the two smaller tiers that sit beneath the flagship Astra model released earlier this month. The company positions the new pair as a way to extend Astra's capabilities to a much wider audience at a fraction of the cost.

Sol is built for demanding work such as software development, while Luna targets high-volume clerical tasks: summarizing documents, extracting information from files, and fielding quick questions where speed matters more than depth.

The headline change is price. API access to the 6-series Sol and Luna costs half of what the 5.6-series versions did, a reduction OpenAI attributes to gains in caching efficiency and inference optimization. Both models are available immediately in ChatGPT Work and Codex for most paid accounts, and through the API. Luna will also appear in the desktop app and for Free and Go tier users, with a gradual rollout to the main ChatGPT app and website underway.

Accuracy is the other pillar of the launch. According to OpenAI, GPT-6 Sol makes roughly half as many factual mistakes as its predecessor on an internal evaluation built from real conversations where users flagged errors, reaching what the company calls Astra-level reliability at a much lower cost. Coding error rates have also dropped, the company says.

Unsurprisingly, the announcement leans into rivalry with Anthropic. OpenAI claims the new models handle tasks substantially better than Anthropic's leading Fable and Opus models, a pointed claim given the timing: Anthropic shipped its own Opus 5.5 release just ninety minutes before OpenAI's announcement.

The back-to-back releases underline how fierce the mid-tier model market has become. With GPT-6 Astra already positioned as the company's most capable system for coding and computer work, the refreshed Sol and Luna are OpenAI's play to make that generation of intelligence, as the company puts it, more efficient and accessible across everyday workloads.