OpenAI, the maker of ChatGPT, released the GPT-5.6 model family on July 9, 2026, in three tiers—Sol, Terra, and Luna—with input pricing from $1 to $5 per million tokens.
The official pricing breaks down by tier: Sol costs $5 per million input tokens and $30 per million output tokens, and OpenAI positions it for frontier reasoning and long-horizon agentic work. Terra runs $2.50 input and $15 output as the balanced everyday option. Luna, at $1 input and $6 output, is the cheapest of the three and the fastest, per OpenAI (vendor-reported).
Prompt caching also becomes more predictable in GPT-5.6: you can set explicit cache breakpoints to mark which parts of a prompt get cached, cached content stays alive for a minimum of 30 minutes, and cache reads keep the 90% cached-input discount. If your prompts carry large fixed blocks, such as system instructions or reference documents, you can now budget for them with more certainty.
Source: OpenAI official announcement, July 9, 2026
