Model comparison
GPT-5.6 Terra vs GPT-5.6 Luna: a flat 10x after July's cuts
GPT-5.6 Terra costs 10x GPT-5.6 Luna on every rate after OpenAI's July 30 price cuts. What each tier is for, and why Codex now sends them different ways.
· Prices as of September 28, 2026
GPT-5.6 Terra
OpenAI · Released July 9, 2026 · Previous generation
The middle tier of GPT-5.6, balancing intelligence and cost, which roughly corresponds to earlier GPT-5 mini models.
GPT-5.6 Terra facts and comparisonsGPT-5.6 Luna
OpenAI · Released July 9, 2026 · Previous generation
The low-cost GPT-5.6 tier for cost-sensitive, high-volume work, which roughly corresponds to earlier GPT-5 nano models.
GPT-5.6 Luna facts and comparisons
The short answer
GPT-5.6 Terra costs exactly 10x GPT-5.6 Luna on every rate, so the example agentic coding session comes to $2.20 against $0.22. OpenAI positions Terra as the tier that balances intelligence and cost and Luna as the one for cost-sensitive, high-volume work. Codex now suggests different successors for them: GPT-6 Sol for Terra and GPT-6 Luna for Luna.
Choose GPT-5.6 Terra if
- You want OpenAI's middle GPT-5.6 tier, which it describes as balancing intelligence and cost.
- You are moving off an earlier GPT-5 mini model, which OpenAI's positioning maps to Terra.
- You plan to follow Codex's suggestion to GPT-6 Sol later and want a GPT-5.6 model on that path now.
Choose GPT-5.6 Luna if
- You run cost-sensitive, high-volume workloads, which OpenAI names as GPT-5.6 Luna's purpose.
- You are replacing an earlier GPT-5 nano model, the role OpenAI says Luna roughly takes over.
- You want 110 sessions a month for $24.20 rather than $242.00.
Side by side
Specs and prices
| Fact | GPT-5.6 Terra | GPT-5.6 Luna |
|---|---|---|
| Maker | OpenAI | OpenAI |
| API model id | gpt-5.6-terra | gpt-5.6-luna |
| Released | July 9, 2026 | July 9, 2026 |
| Status | Previous generation | Previous generation |
| Context window | 1.05M tokens | 1.05M tokens |
| Max output | 128K tokens | 128K tokens |
| Open weights | No | No |
| Input, per 1M tokens | $2 | $0.20 |
| Cache hit, per 1M | $0.20 | $0.02 |
| Cache write, per 1M | $2.50 | $0.25 |
| Output, per 1M tokens | $12 | $1.20 |
| Runs in | Codex, Cursor, OpenCode, OpenRouter, and GitHub Copilot | Codex, Cursor, OpenCode, OpenRouter, and GitHub Copilot |
Standard API rates in US dollars, as published by each maker on September 28, 2026. Batch and priority tiers, taxes, and subscription plans are not included. GPT-5.6 Terra: Requests over 272K input tokens cost 2x for input and cache and 1.5x for output, for the whole request. GPT-5.6 Luna: Requests over 272K input tokens cost 2x for input and cache and 1.5x for output, for the whole request.
Cost
What the same work costs
The same token counts, priced at each model’s published rates. Cached tokens are billed at each maker’s cache prices, so the session shows what caching is worth on each model.
Real sessions differ: the two models count the same code as different numbers of tokens, and reasoning settings change how much each one writes. Your own history is the real test.
| Workload | GPT-5.6 Terra | GPT-5.6 Luna |
|---|---|---|
| Agentic coding session, 100K input, 400K written to cache, 2M read from cache, 50K output | $2.20 | $0.22 |
| Large one-off review, 150K input with no cache hits, 10K output | $0.42 | $0.04 |
| Output-heavy generation, 30K input, 80K output | $1.02 | $0.10 |
| A month of sessions, 110 sessions: 5 a day, 22 working days | $242.00 | $24.20 |
| Where the session’s cost goes | ||
| Cache writes | $1.00 | $0.10 |
| Cache reads | $0.40 | $0.04 |
| Uncached input | $0.20 | $0.02 |
| Output | $0.60 | $0.06 |
- caching saves on the session with GPT-5.6 Terra (61%)
- $3.40
- caching saves on the session with GPT-5.6 Luna (61%)
- $0.34
Is GPT-5.6 Terra exactly 10x the price of GPT-5.6 Luna?
Per token, yes. GPT-5.6 Terra lists input at $2 and output at $12 per million tokens, cache writes at $2.50, and cache hits at $0.20. GPT-5.6 Luna charges $0.20, $1.20, $0.25, and $0.02 for the same four. Every rate is 10x apart.
Workload totals round to the cent, which shows at Luna's scale. The session reads $2.20 against $0.22, an even 10x. The uncached review reads $0.42 against $0.04, or 10.5x, and the output-heavy generation $1.02 against $0.10, or 10.2x. The per-token rates, not the rounded totals, carry the real ratio.
The session splits identically on both: 45% cache writes, 27% output, 18% cache reads, and 9% fresh input. Caching saves 61% of the uncached cost on each, $3.40 on Terra and $0.34 on Luna.
What the July 30 price cuts did to the GPT-5.6 lineup
OpenAI launched GPT-5.6 on July 9, 2026, and cut prices three weeks later, on July 30: Terra by 20% and Luna by 80%. Because Luna's cut was much deeper, the gap between the two tiers widened. The rates on this page are the post-cut rates.
OpenAI's own descriptions map the tiers to older names. It places Terra roughly where earlier GPT-5 mini models sat and Luna roughly where earlier GPT-5 nano models sat. It calls Terra the model that balances intelligence and cost, and Luna the one optimized for cost-sensitive workloads.
Both share the family traits OpenAI claims for GPT-5.6, such as token efficiency and better frontend design. Those are family-wide claims, so they don't tell you where Luna's limits fall relative to Terra on your tasks.
Codex points Terra and Luna to different GPT-6 models
Codex suggests moving from GPT-5.6 Terra to GPT-6 Sol and from GPT-5.6 Luna to GPT-6 Luna. Both GPT-5.6 models stay available in the API. At the next generation, Terra's successor is the mid-priced GPT-6 model and Luna's is the cheapest.
The two share every limit and rule: a 1.05M context window, 128K of output, cache writes at 1.25x input, and 2x input and cache plus 1.5x output on requests over 272K input tokens. Both run in Codex, Cursor, OpenCode, OpenRouter, and GitHub Copilot.
At a 10x spread, routing matters more than the headline price. Sending narrow, repeatable jobs to Luna and keeping broader ones on Terra is the split OpenAI's positioning implies. EveryToken shows your local per-model spending across Codex, Cursor, and OpenCode, so you can see how much of it already runs on each tier.
Prompt caching
How OpenAI bills cached tokens
OpenAI
Prompt caching is on by default. From GPT-5.6 on you can also mark up to four explicit cache breakpoints, while GPT-5.5 and earlier cache automatically only.
From GPT-5.6 on, a cache write costs 1.25x the uncached input price and a cache hit costs 0.1x. GPT-5.5 and earlier add no charge for writing the cache: written tokens are billed as ordinary input, and a hit costs 0.1x on the models compared here.
On GPT-5.6 and later, a cached prefix stays reusable for at least 30 minutes after its last use, and caching starts at 1,024 input tokens.
Source: OpenAI: Prompt caching
Your own numbers
See what GPT-5.6 Terra and GPT-5.6 Luna really cost you.
everyaitoken reads your Codex, Cursor, OpenCode, and OpenRouter history on your Mac and prices every request at API rates, with what caching saved or cost. $9 once.
FAQ
Questions
How much cheaper is GPT-5.6 Luna than GPT-5.6 Terra?
Luna's rates are a tenth of Terra's across input, output, and both cache prices. A month of 110 example sessions costs $24.20 on Luna and $242.00 on Terra.
Did OpenAI cut GPT-5.6 prices?
Yes. On July 30, 2026, OpenAI cut GPT-5.6 Terra's price by 20% and GPT-5.6 Luna's by 80%.
What should I move to from GPT-5.6 Terra or GPT-5.6 Luna?
Codex suggests GPT-6 Sol for Terra users and GPT-6 Luna for Luna users. Both GPT-5.6 models stay available in the API if you are not ready to switch.
Are GPT-5.6 Terra and GPT-5.6 Luna the same as GPT-5 mini and nano?
No, they are separate GPT-5.6 models. OpenAI's positioning places Terra roughly where earlier GPT-5 mini models sat and Luna roughly where earlier GPT-5 nano models sat.