Introducing GPT-6 Sol and Luna
GPT-6 Sol and Luna offer different capability-cost balances, giving builders two model tiers to test against the quality, latency, and budget needs of agent workloads.
OpenAI introduces **GPT-6 Sol** and **GPT-6 Luna** as two models with different balances of capability and cost for everyday work.
Agent builders should evaluate both on their own tasks and route work by required capability and acceptable spend rather than treating them as interchangeable defaults.
OpenAI introduces **GPT-6 Sol** and **GPT-6 Luna** as two models with different balances of capability and cost for everyday work. Agent builders should evaluate both on their own tasks and route work by required capability and acceptable spend rather than treating them as interchangeable defaults. The supplied material gives no prices, benchmarks, context limits, or task-specific guidance, so it does not establish where the crossover between the models lies.
This formalizes capability-versus-cost routing within OpenAI’s GPT-6 family, but does not provide the measurements needed to assign Sol or Luna to particular workloads. It therefore expands the evaluation matrix rather than establishing a new default: builders still need matched trials for quality, latency, reliability, and total cost.