openai / gpt-5-6-luna

Overview

GPT 5.6 Luna is OpenAI’s speed-first tier in the GPT‑5.6 model family. It takes your written instructions and optional images, then produces a text answer that explains what it sees and what to do next. It works well for high-volume help, fast analysis, and visual Q&A where you need clear text output.

What you can build

• Screenshot Q&A for apps, dashboards, and error dialogs
• Product photo review for defects, parts, or basic attributes
• Visual document triage, like “what is this page about”
• Image captioning and alt-text drafts for accessibility
• Simple diagram explanations and step-by-step descriptions
• Chat-style support flows that keep session context across turns

Inputs

• A written prompt that explains what you want, as plain text.
• Optional one or more image files to analyze. If you provide multiple images, separate them with a semicolon.
• Optional user identifier, as a string or number. Use it when you want the model to keep chat history per user.
• Optional session identifier, as a string or number. Use it to separate multiple conversations for the same user.
• Optional system-level instructions, as plain text. Keep this under 5,000 characters.
• Optional reasoning intensity selection. Choose from no extra reasoning through deeper levels up to Max.
• Optional web search toggle. Turn it on when you need fresh public information.
• Optional verbosity choice. Pick low, medium, or high detail in the response.

Outputs

• A text response that answers your prompt.
• When you provide images, the response includes visual observations tied to your question.
• When you provide user and session identifiers, the model can keep conversational continuity across messages in the same session.

Limitations

• This runner is built for image understanding with text output. It does not generate images, audio, or video.
• Low-quality inputs reduce accuracy. Blurry photos, glare, tiny text, heavy compression, and busy layouts can cause mistakes.
• Safety systems can block or restrict some requests, especially in higher-risk areas.
• Some requests may return less detail than you asked for when the model must follow safety rules.

Safety & compliance

• GPT 5.6 Luna uses layered safeguards from OpenAI.
• Expect extra checks for sensitive topics, including certain biology and cybersecurity content.
• Don’t use it to request wrongdoing, evasion, or instructions that could cause harm.
• If you process personal or confidential images, get proper rights and consent first.