What it does
It is in limited preview. OpenAI claims 150-millisecond responses, but pricing, candidate limits and customization remain undisclosed.
Maker-reported (not independently measured by JevMade): OpenAI reports 150 ms responses, compared with 1.6 seconds for GPT-6 Luna.
- Primitives
- Not stated
- Added
Source checked 2026-09-29 — opened the primary source directly.