OpenAI · 2026-06-26 · seismic
GPT-5.6 — OpenAI previews Sol, Terra, and Luna tiers
OpenAI announced GPT-5.6 with three named tiers — Sol (flagship), Terra (balanced), and Luna (cheap and fast) — adding new max and ultra reasoning modes. Access starts as a limited preview for trusted partners.
OpenAI's new generation splits into three named tiers and adds an ultra mode that wires subagents into the flagship model.
Quick facts
| Maker | OpenAI |
|---|---|
| Tiers | Sol (flagship), Terra (balanced), Luna (cheap and fast) |
| New reasoning modes | max effort (Sol) and ultra mode using subagents |
| Focus areas | coding, biology, cybersecurity |
| Availability | Limited preview to trusted partners; broader rollout in the coming weeks |
| Price (Sol) | $5 in / $30 out per 1M tokens |
| Announced | June 26, 2026 |
Pricing
| Sol input | $5.00 / 1M tokens |
|---|---|
| Sol output | $30.00 / 1M tokens |
| Terra input | $2.50 / 1M tokens |
| Terra output | $15.00 / 1M tokens |
| Luna input | $1.00 / 1M tokens |
| Luna output | $6.00 / 1M tokens |
| Cache reads | 90% off input rate |
What is it?
GPT-5.6 is OpenAI's next generation of general-purpose models, arriving roughly two months after GPT-5.5. The release ships as three durable tiers — Sol for the hardest work, Terra as the everyday default, and Luna as the cheap and fast option — instead of one undifferentiated model. The 5.6 number labels the generation; the names label capability tiers that can each move on their own cadence.
How does it work?
Sol gets two new reasoning options. Max effort lets the flagship think longer on a single hard problem, while ultra mode goes beyond a single agent by spinning up subagents that tackle parts of the task in parallel. OpenAI also says the family was developed with its strongest safeguards to date, and was audited for cross-persona reward signal leakage before training. All three sizes show measurable jumps in coding, biology, and cybersecurity work over GPT-5.5.
Why does it matter?
Splitting GPT-5.6 into Sol, Terra, and Luna gives developers a clean way to choose between cost and depth without juggling preview names — and prices it transparently from day one ($5/$30 for Sol, $1/$6 for Luna per million tokens). For agent builders, ultra mode is the bigger story: the flagship can now orchestrate subagents itself, which moves logic that used to sit in tools like LangGraph back inside the model.
Who is it for?
API users, agent builders, ChatGPT power users
Frequently asked questions
- How much does GPT-5.6 cost?
- GPT-5.6 prices per million tokens are $5 input / $30 output for Sol, $2.50 / $15 for Terra, and $1 / $6 for Luna. Cache writes cost 1.25x the uncached input rate, and cache reads get a 90% discount. The numbers were published in OpenAI's preview post on June 26, 2026.
- Who can use GPT-5.6 right now?
- Access to GPT-5.6 is limited at launch. OpenAI says it is starting with a small group of trusted partners and organizations whose participation has been shared with the U.S. government. Sol, Terra, and Luna are planned for broader release to ChatGPT and the API in the coming weeks.
- What do Sol, Terra, and Luna actually mean?
- Sol, Terra, and Luna are durable capability tiers that can each advance on their own cadence, while the 5.6 number names the generation. Sol is the flagship for the hardest reasoning work, Terra is the balanced everyday model, and Luna is the fast and affordable option. OpenAI describes them as named after celestial bodies.
- What are max and ultra reasoning modes?
- GPT-5.6 Sol adds two new effort settings. Max reasoning gives Sol the longest think time on a single hard problem, and ultra mode goes beyond a single agent by orchestrating subagents to work in parallel. OpenAI says the modes target deep coding, biology, and cybersecurity tasks where ordinary reasoning runs out of room.
Try it
openai.com/index/previewing-gpt-5-6-sol