← Models

GPT-5.6 Luna

OpenAI · GPT-5.6

OpenAI's lowest-cost GPT-5.6 model for high-volume agents, coding support, extraction, and efficient professional workflows.

Part of GPT family · Other versions: GPT-5.6 Terra , GPT-5.6 Sol , GPT-5.5 Pro , GPT-5.5 , GPT-Rosalind , GPT-5.4 mini , GPT-5.4 nano , GPT-5.4 Pro , GPT-5.4 , GPT-5.3 , GPT-5.3-Codex , GPT-5.2-Codex , GPT-5.2 Pro , GPT-5.2 , GPT-5-Codex , GPT-5 mini , GPT-5 nano , GPT-5 , GPT-4.1 , GPT-4o mini Transcribe , GPT-4o mini TTS , GPT-4o Transcribe , Computer Use Preview , GPT-4o mini , GPT-4o
Type
language
Context
1M tokens
Max Output
128K tokens
Status
current
Input
$1/1M tok
Output
$6/1M tok
API Access
Yes
License
proprietary
reasoning coding agentic tool-use computer-use long-context high-volume cost-efficient
Released July 2026 · Updated July 10, 2026

Overview

Freshness note: Model capabilities, limits, pricing, and availability can change quickly. This profile is a point-in-time snapshot last verified on July 10, 2026.

GPT-5.6 Luna is OpenAI’s fastest and most affordable GPT-5.6 tier. It is intended for cost-sensitive and high-volume workloads that still benefit from the current generation’s reasoning, tool use, long context, and agent features. OpenAI compares its lineup position to the nano tier used in earlier GPT-5 families, but Luna’s documented 1.05M-token context and 128K output ceiling are much larger than a typical small-model profile.

Luna became generally available on July 9, 2026 alongside GPT-5.6 Sol and GPT-5.6 Terra.

Capabilities

Luna supports reasoning, coding assistance, structured extraction, document processing, tool calling, and computer-use style workflows. It also inherits the GPT-5.6 generation’s Programmatic Tool Calling, persisted reasoning, explicit prompt-cache controls, and multi-agent beta support through the Responses API.

Reasoning effort ranges from none to max. High effort on a low-cost model is not automatically the cheapest route, however: a shorter Terra run may cost less than a long Luna run. Route by measured task cost and success rate rather than model price alone.

Technical Details

OpenAI’s API docs list:

  • Model ID: gpt-5.6-luna
  • Context window: 1,050,000 tokens
  • Maximum output: 128,000 tokens
  • Knowledge cutoff: February 16, 2026
  • Input: text and images
  • Output: text

Streaming, function calling, and structured outputs are supported. The current GPT-5.6 tool surface covers search, files, code interpreter, hosted shell, patching, skills, computer use, MCP, and tool search. Fine-tuning is not currently supported.

Like the other GPT-5.6 tiers, Luna can use reasoning.mode: "pro"; OpenAI does not expose a separate Luna Pro model slug.

Pricing & Access

Published standard API pricing per 1M tokens is:

  • Input: $1.00
  • Cached input: $0.10
  • Output: $6.00

Prompts above 272K input tokens are priced at 2x input and 1.5x output for the full request. Explicit cache writes cost 1.25x the uncached input rate, so cache economics depend on reuse frequency rather than the read price alone.

Luna is available through the OpenAI API, ChatGPT Work, and Codex where the relevant plan and workspace expose GPT-5.6 model selection.

Best Use Cases

Choose Luna for classification, extraction, routine transformation, high-volume drafting, coding support, search-result processing, and parallel subagent work where throughput and cost matter more than maximum single-agent quality.

Escalate to Terra when Luna needs too many retries or too much reasoning, and to Sol when the task has expensive failure modes, complex ambiguity, or a high quality bar for the final artifact.

Comparisons

  • GPT-5.6 Terra (OpenAI): Stronger balanced tier for production work that needs more consistent reasoning.
  • GPT-5.6 Sol (OpenAI): Flagship route for the hardest coding and professional workflows.
  • GPT-5.4 nano (OpenAI): Cheaper earlier-generation nano route for simpler workloads and pinned integrations.
  • Gemini 3.5 Flash (Google): Fast competing multimodal route with Google-native tools and product integration.