OpenAI
PreviewGPT-5.6 pricing
OpenAI's June 2026 model family, split into three tiers: Sol for the hardest problems, Terra for high-volume business tasks, and Luna for fast everyday work.
Input
$5.00/ 1M tok
Output
$30.00/ 1M tok
- Context window
- 1.5M
- Max output
- 128K
- Cached input
- -
- Verified
- 2026-07-01
GPT-5.6 tiers
| Tier | Input / 1M | Output / 1M | Cached in / 1M |
|---|---|---|---|
Sol Hardest problems: complex coding, security research | $5.00 | $30.00 | - |
Terra High-volume business: support, internal tools, docs | $2.50 | $15.00 | - |
Luna Fast everyday: summarisation, drafting, automation | $1.00 | $6.00 | - |
Headline figures above are for the Sol tier. GPT-5.6 launched in limited preview on 26 June 2026 (~20 organisations); wider availability is expected in the following weeks.
GPT-5.6 is OpenAI's June 2026 release and, unlike earlier point releases, ships as a three-model family rather than a single flagship. Sol targets the hardest problems – complex coding and security research – at $5 input / $30 output per 1M tokens. Terra is the high-volume business tier at $2.50 / $15, and Luna is a new low-cost everyday tier at $1 / $6, aimed at summarisation, drafting, and routine automation.
The context window is reported at roughly 1.5M tokens (up from 1M on GPT-5.5), though OpenAI has not published an official figure. GPT-5.6 also tightens prompt caching: cache writes bill at 1.25x the uncached input rate, while cache reads keep the 90% discount, with explicit cache breakpoints and a 30-minute minimum cache life for more predictable savings on repeated prefixes.
Because the family spans a 5x price range from Luna to Sol, model selection is where the cost lever lives: route latency-tolerant, high-volume work to Luna or Terra and reserve Sol for the genuinely hard requests. As of launch the models are limited to a small preview cohort, with general availability expected shortly after.
Estimate LLM costs before you send
Paste your prompt into the Calcis estimator to see token counts and per-request cost across every tracked model, then compare GPT-5.6 against them side by side.
Frequently asked
- How much does GPT-5.6 cost per 1M tokens?
- It depends on the tier. Sol is $5 input / $30 output, Terra is $2.50 / $15, and Luna is $1 / $6, all per 1M tokens.
- What is the difference between Sol, Terra, and Luna?
- Sol is tuned for the hardest problems (complex coding, security research), Terra for high-volume business tasks (support, internal tools, document analysis), and Luna for faster, lower-cost everyday work (summarisation, drafting, routine automation).
- What is the GPT-5.6 context window?
- Reported at approximately 1.5M tokens, up from the 1M GPT-5.5 shipped with. OpenAI has not published an official figure at launch.
- How does prompt caching work on GPT-5.6?
- Cache writes bill at 1.25x the model's uncached input rate; cache reads receive the standard 90% cached-input discount. GPT-5.6 adds explicit cache breakpoints and a 30-minute minimum cache life.
- Is GPT-5.6 generally available?
- Not at launch. It debuted on 26 June 2026 for a limited preview cohort of roughly 20 organisations, with wider availability expected in the following weeks.
Pricing verified 2026-07-01 from the provider's rate card. These figures are informational and not yet wired into the Calcis estimator or billing.