GPT-5.6 Luna Pro

OpenAI's June 30, 2026 GeneBench-Pro paper is the first-party source that explicitly names the GPT-5.6 Luna Pro variant. The paper introduces a 129-problem benchmark spanning genomics, quantitative biology, and translational medicine, designed to test whether AI agents can make judgment-heavy analytical decisions on me Within the GeneBench-Pro results, GPT-5.6 Luna Pro appears as a distinct configuration alongside the standard Luna, Terra, and Sol tiers, marking the first time an official OpenAI document has listed multiple Pro-tier configurations for a single generation. The paper positions GPT-5.6 Luna Pro (and its sibling Terra Pr

TechTimes' July 2, 2026 report provides concrete performance data for GPT-5.6 Luna Pro drawn from OpenAI's GeneBench-Pro results table. According to the report, GPT-5.6 Luna achieved a 16.5% pass rate at its standard max reasoning setting, which climbed to 23.6% with "Pro (Extended)" mode enabled — a +7.1 point gain, t The report frames GPT-5.6 Luna Pro's appearance in an official OpenAI benchmark as a disclosure of previously unannounced Pro-tier variants, noting that GPT-5.6 Sol's standard configuration cleared 28.7% of GeneBench-Pro problems while its Pro mode reached 31.5%. The piece also extrapolates about a potential ChatGPT Pr