What changed
OpenAI’s current API catalog identifies GPT-6 Astra for complex reasoning and coding, GPT-6.1 Sol for a balance of capability and cost, and GPT-6 Luna for focused, high-volume work. These are the provider’s positioning statements.
The everyday takeaway
Pick a model around the task and the amount of checking it needs. A short rewrite and a long debugging session do not necessarily justify the same configuration. App access and API access are different products.
What we have verified
OpenAI’s API changelog dates GPT-6.1 Sol’s release to September 29. The catalog establishes the current lineup; it does not establish the same availability in every app or subscription. We have not run a controlled comparison of these three models for this briefing.
A second perspective
A measured look at the updated middle model. Treat the results as evidence about the evaluator’s tasks and configurations.