GPT-5.6 Luna
Language modelPrevious generationCard summarizedGPT-5.6 model optimized for cost-sensitive workloads
Active (legacy)● Re-checked 12 Oct 2026 · 84 facts · 0 corrections · 1 unconfirmedSource: GPT-5.6 System Card (82 pp)
Use it via API
Where to call it
- OpenAI API
gpt-5.6-luna
Supports
- Image input
- Streaming
- Structured outputs
- Reasoning control
- Tool calling
- Web search
- Prompt caching
- Batch API
Good for
- Fastest and most cost-efficient GPT-5.6 model
- Beats GPT-5.5 on HealthBench Professional (55.7 vs 51.8) despite its lower cost
- Lowest indirect prompt-injection success rate of the family (2.94%)
Watch out for
Overview
GPT-5.6 Luna
GPT-5.6 model optimized for cost-sensitive workloads
- Cost
- $0.20 in
- Context
- 1.05M
- Input / output
- $0.20 / $1.20
- Max output
- 128K
- Knows up to
- February 2026
- Released
- 9 Jul 2026
The fastest and most cost-efficient GPT-5.6 model. Together with Terra, it is the first time smaller and faster members of a model family received a High designation: High in cybersecurity (below Critical) and in biology and chemistry, below High in AI self-improvement. It is protected by topical classifiers and a safety reasoner but not the activation classifiers used for Sol and Terra. Most alignment and external evaluations in the card cover Sol only. p.2–3, p.35, p.47, p.75–76
gpt-5.6-luna- Input / output per 1M
- $0.20 / $1.20
- Context window
- 1.05M
- Max output
- 128K
- Knows up to
- February 2026
- Reasoning
- Configurable (effort none → max)
- Weights
- Closed
About this card. A model card is the maker's own account of a model: what it's for, how it was trained and tested, and where it falls short. This page condenses OpenAI's GPT-5.6 System Card. Every claim carries its source page.