OpenAI
GPT-5.6 Luna: price, context window, and caching
The low-cost GPT-5.6 tier for cost-sensitive, high-volume work, which roughly corresponds to earlier GPT-5 nano models.
Released July 9, 2026 · Prices as of September 28, 2026
In OpenAI’s words
“GPT-5.6 model optimized for cost-sensitive workloads”
Facts
Specs and prices
| Fact | GPT-5.6 Luna |
|---|---|
| Maker | OpenAI |
| API model id | gpt-5.6-luna |
| Released | July 9, 2026 |
| Status | Previous generation |
| Context window | 1.05M tokens |
| Max output | 128K tokens |
| Open weights | No |
| Input, per 1M tokens | $0.20 |
| Cache hit, per 1M | $0.02 |
| Cache write, per 1M | $0.25 |
| Output, per 1M tokens | $1.20 |
| Runs in | Codex, Cursor, OpenCode, OpenRouter, and GitHub Copilot |
Standard API rates in US dollars, as published by the maker on September 28, 2026. Batch and priority tiers, taxes, and subscription plans are not included. GPT-5.6 Luna: Requests over 272K input tokens cost 2x for input and cache and 1.5x for output, for the whole request.
Good to know
- Codex suggests moving to GPT-6 Luna. It stays available in the API.
- OpenAI cut its price by 80% on July 30, 2026.
Cost
What typical work costs
Example token counts at GPT-5.6 Luna’s published rates. On the agentic session, caching saves $0.34 against billing every token as ordinary input.
| Agentic coding session, 100K input, 400K written to cache, 2M read from cache, 50K output | $0.22 |
|---|---|
| Large one-off review, 150K input with no cache hits, 10K output | $0.04 |
| Output-heavy generation, 30K input, 80K output | $0.10 |
| A month of sessions, 110 sessions: 5 a day, 22 working days | $24.20 |
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
GPT-5.6 Luna compared
Claude Haiku 4.5 vs GPT-5.6 Luna
After an 80% price cut, GPT-5.6 Luna costs a fifth of Claude Haiku 4.5 for input. A month of cached coding sessions: $24.20 against $132.00.
DeepSeek-V4.1-Flash vs GPT-5.6 Luna
DeepSeek-V4.1-Flash and GPT-5.6 Luna both price a cached coding session at $0.22, from opposite ends: Luna's cheaper input, DeepSeek's cheaper cache hits.
GPT-5.6 Luna vs Gemini 3.5 Flash-Lite
Gemini 3.5 Flash-Lite costs 1.5x as much as GPT-5.6 Luna on a cached coding session, $0.34 against $0.22, and about twice as much on output-heavy work.
GPT-5.6 Terra vs GPT-5.6 Luna
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.
GPT-6 Luna vs GPT-5.6 Luna
GPT-6 Luna halves the input price of GPT-5.6 Luna, which already had an 80% cut, and trims output further. What the move saves, and what Codex suggests.
Guides
When prompt caching costs more than it saves
Cache writes cost more than ordinary input on Claude and newer GPT models. Worked examples show when a write never pays back and caching adds cost.
5-minute vs 1-hour cache writes: which pays off?
Anthropic's 1-hour cache write costs 2x input, the 5-minute write 1.25x. When the longer lifetime saves money depends on the gaps between your requests.
What is API-equivalent cost?
API-equivalent cost is what your AI coding tokens would cost at the provider's published API rates. What it measures, how it is computed, and what it is not.
Your own numbers
See what GPT-5.6 Luna really costs 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. $9.99 once.