Text
GPT-6 Luna in Twelver: the new Luna at half the price of GPT-5.6 Luna
Luna has moved to the GPT-6 generation. If you already call gpt, there is nothing to change: your requests go to the new model and cost less.
What’s new
- Input at half the price, output 2.4x cheaper than GPT-5.6 Luna.
- Knowledge up to May 2026 — three months fresher than its predecessor.
- A 1.05M-token context and answers up to 128K tokens, as before.
- Reasoning switches off completely — simple requests spend no reasoning tokens at all.
- Same API name, gpt: integrations move to the new model with no code changes.
At a glance
| Detail | Value |
|---|---|
| Model | GPT-6 Luna (OpenAI) |
| Replaces | GPT-5.6 Luna |
| Context | 1.05M tokens |
| Max answer | 128K tokens |
| Knowledge | Up to May 2026 |
| Reasoning | From off up to xhigh |
| Where in Twelver | OpenAI-compatible API |
| API name | gpt (flagship GPT-6 Astra is gpt-max) |
Frequently asked questions
How does GPT-6 Luna differ from GPT-5.6 Luna?
It is newer, knows events up to May 2026 and costs noticeably less: half on input and 2.4x less on output. Context and maximum answer length are the same. Both GPT models are covered in the GPT chapter.
Do I need to change anything if I used GPT-5.6 Luna?
No. The API model name is still gpt, and from 25 September 2026 it answers on GPT-6 Luna.
Do I need an OpenAI account?
No. Requests go through your Twelver account over the OpenAI-compatible API, so it works from your own code or from editors such as Cursor and Cline.
How much does GPT-6 Luna cost?
Current rates are on the API page. OpenAI bills prompts over 272K tokens at a higher rate, and we bill them the same way.
Try it
Open in Twelver chatRate your experience