Enterprise customers told OpenAI its models were too expensive to run at scale. OpenAI, to its credit, actually listened.
The company launched the GPT-5.6 model family on July 9, featuring three distinct variants designed to give businesses options that don’t require a second mortgage on their cloud budgets. CEO Sam Altman framed the release as a direct response to enterprise concerns about runaway AI inference costs.
Three tiers, one message: costs matter
The GPT-5.6 lineup breaks down into Sol, Terra, and Luna.
Sol is the flagship. It’s priced at $5 per million input tokens and $30 per million output tokens. The headline number here is a 54% improvement in token efficiency for agentic coding tasks compared to rival models.
