OpenAI's GPT-5.6 is here — but you probably can't use it yet
OpenAI launched GPT-5.6 Sol, Terra, and Luna as a gated preview. The three-tier pricing and the access gate matter more to your business than the benchmarks.
OpenAI shipped its next frontier model, GPT-5.6, on June 26 — and most businesses can't touch it. It launched as a limited preview for a small group of partners "shared with the government," with general availability promised "in the coming weeks." The useful story for an operator isn't the flagship's benchmarks. It's the price sheet and the waitlist.
What actually happened
OpenAI released GPT-5.6 in three tiers — Sol (flagship), Terra (balanced, everyday), and Luna (fast and cheap) — across ChatGPT and Codex, per 9to5Mac. Access started as a limited preview for trusted partners whose participation was shared with the government, with broad availability expected "in the coming weeks," per VentureBeat. Sol adds a max reasoning effort and an ultra mode that spins up subagents to work in parallel.
The prices tell the real story. Per million tokens: Sol runs $5 in / $30 out, Terra $2.50 / $15, Luna $1 / $6. That's a 5x spread from top to bottom — for models in the same family, released the same day.
Why it matters for your business
Two takeaways, neither about which model wins a benchmark.
First: the frontier is a waitlist now. If your operation is architected to depend on the newest, most powerful model the day it drops, you've built on something you can't reliably get. The models you can actually deploy today — generally available, documented, rate-limited but available — are the ones worth building on. Treat day-one frontier access as a bonus, not a foundation.
Second: the tier you pick is a cost decision, not a status one. Most real business work — classify this ticket, draft this reply, extract these fields — does not need a $30-output flagship. It runs fine on Luna or Terra at a fraction of the price. Reaching for the top model on every call is how a working automation quietly turns into a five-figure invoice. And ultra mode's subagents, run unwatched, burn tokens fast.
We build so the model is a swappable dial, not a load-bearing wall — match the tier to the task, and change it when the price sheet does.
Key takeaways
- OpenAI launched GPT-5.6 in three tiers — Sol (flagship), Terra (balanced), Luna (fast/cheap) — on June 26, in ChatGPT and Codex
- Access is a limited preview "shared with the government"; general availability is promised "in the coming weeks"
- Pricing spans 5x: Sol $5/$30, Terra $2.50/$15, Luna $1/$6 per million tokens
- Operator moves: build on models you can actually get today, and match the tier to the task instead of defaulting to the flagship
Not sure which model your automation should actually be calling? We build systems that route each task to the cheapest model that does the job — and swap tiers as prices move. Start here or tell us what you're running.
Sources: 9to5Mac, VentureBeat.
- #gpt-5-6
- #openai
- #llm
- #ai-costs
Tommy Rush — Founder, Rush Commerce
Operator turned builder. 15+ years running operations — now shipping the systems businesses run on. More
Get The Rush Report weekly — one email, zero fluff.
Keep reading
Claude Sonnet 5 is cheaper on paper — why your bill might not be
Anthropic launched Claude Sonnet 5 as a lower-cost agentic model, but a new tokenizer and an August price step-up mean a flat rate card isn't a flat bill. The production lesson for anyone running AI in their operation.
Read itHow much does custom software cost? A real 2026 breakdown
Straight numbers from someone who builds it: what websites, internal tools, apps, and automations actually cost in 2026 — and what drives the price up or down.
Read it