OpenAI GPT-5.6 Pricing Update Cuts Terra and Luna Costs, Leaves Sol Unchanged

OpenAI has lowered the prices for the Terra and Luna tiers within its GPT-5.6 model family, while keeping the standard pricing for Sol unchanged. For API users, this means there are now lower-cost model options that can help reduce routine inference spend, but organizations using Sol should not expect a broad temporary reduction in its standard token rates.

The price reductions are significant: Luna pricing is now 80% lower and Terra pricing is 20% lower than before. Additionally, OpenAI has introduced a Fast mode for Sol, promising up to 2.5 times faster speeds at twice the cost; however, this is not a price reduction but rather a throughput option.

When GPT-5.6 launched, Sol was priced at $5 per million input tokens and $30 per million output tokens. Terra was at $2.50 for input and $15 for output, while Luna was at $1 for input and $6 for output. The updated pricing maintains these figures for Sol, with Terra and Luna seeing the price cuts.

OpenAI emphasizes that the changes underscore a tiered approach to model selection, with Sol remaining the flagship tier and Terra and Luna offering more affordable options. The update supports distinguishing between the need for Sol's speed and the cost-effectiveness of Terra and Luna, encouraging teams to evaluate workload requirements rather than assuming a uniform price change across the model family.

For businesses utilizing AI-powered services, this pricing update highlights the importance of model governance. With lower token costs, there's an opportunity for improved cost efficiency in experimentation and scaling. However, it's crucial to maintain a clear inventory of which applications use which tier and for what purposes to manage spending effectively.

Sol Fast mode adds another layer of complexity, acting as a premium option for speed rather than a price cut. Organizations should therefore treat it as a separate consumption consideration.