OpenAI Adds GPT-6 Sol and Luna With 50% Lower API Prices

OpenAI expanded its GPT-6 family with Sol and Luna, which it says were trained with methods similar to Astra and carry advances in professional work, factuality, coding, computer use, and alignment into faster, more affordable models. [1] API prices are 50% below GPT-5.6 promotional pricing: Sol costs $2 per million input tokens and $10 per million output tokens, while Luna costs $0.10 and $0.50, respectively. [1] OpenAI continues to position GPT-6 Astra as its best model across the board. [1]

GPT-6 Sol and Luna are generally available on Amazon Bedrock through an inference engine that AWS describes as built for performance, security, and reliability at scale. [2] AWS positions Sol for recurring, demanding development and operations tasks, while Luna targets focused, repeatable work performed at high volume. [2] OpenAI reports that Sol at xhigh effort outperforms Claude Opus 5 at max effort on AutomationBench while costing 9% as much per task. [1] TechCrunch reports that OpenAI’s internal factuality evaluation found Sol made about half as many mistakes as its predecessor and reached Astra-level reliability at lower cost. [3]

Sol and Luna make the GPT-6 lineup a tiered deployment choice: Astra for maximum quality, Sol for recurring complex work, and Luna for focused volume. OpenAI’s AutomationBench comparison puts Sol ahead of Claude Opus 5 at a fraction of the stated task cost, but the evidence is vendor-reported rather than an independent evaluation. Bedrock availability provides organizations with another production route, while lower token prices alter the economics of repeated calls. What the sources don't address: how Sol and Luna perform under independent testing across real production workloads.