OpenAI launches GPT-6 Sol and Luna minutes after Claude Opus 5.5

OpenAI released GPT-6 Sol and GPT-6 Luna on Tuesday, minutes after Anthropic launched Claude Opus 5.5 as its new flagship. Sol and Luna sit below GPT-6 Astra, which OpenAI called the most intelligent and aligned model in the world when it launched on September 3. OpenAI presents Astra as its top model for the hardest jobs, while Sol and Luna are cheaper and faster options for everyday work. The article compares the lineup with Anthropic's strategy, describing Astra as equivalent to Fable, Sol to Opus, Terra to Sonnet, and Luna to Haiku when it is upgraded. OpenAI cut the API prices for both models by 50% compared with GPT-5.6's promotional rates per token. Sol costs $2 per million input tokens and $10 per million output tokens, down from $4 and $20. Luna costs $0.10 per million input tokens and $0.50 per million output tokens, down from $0.20 and $1.20. The article states that OpenAI is the cheaper offer per tier compared with Anthropic. On AutomationBench, a Zapier benchmark that tests whether an AI agent can complete a business workflow using 47 tools across sales, marketing, operations, support, finance, and human resources, Sol scored 33.2% at its highest reasoning setting, at a cost of $0.27 per task. Claude Opus 5 scored 26.9% at its highest setting and cost more than 11 times as much per task, according to OpenAI's own figures. Sol also edged out Claude Fable 5.1, described as Anthropic's pricier flagship, on the same test. On Agents' Last Exam, which measures long, economically valuable work across 55 sub-industries as a percentage completed, Sol scored 56.4% at its highest effort. OpenAI says that score beats Claude Opus 5's best result at a 60% lower cost per task. OpenAI's reasoning-effort setting makes the model spend more time and computing power checking its answers, which tends to increase both accuracy and cost. On OSWorld 2.0, a benchmark for agents that click, type, and navigate software, Sol nearly matched Claude Opus 5's medium-effort score, at 60.5% versus 60.3%, while costing about 80% less. GPT-6 Sol and Luna are live in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu subscribers. Luna is also available to free and Go users through the desktop app. Neither model is yet available in the standard ChatGPT app, and OpenAI says it is rolling them out gradually throughout the day to maintain service stability.