GPT-5.6 Luna

Available · Language, Multimodal, Reasoning

GPT-5.6 Luna is the smallest tier of OpenAI's GPT-5.6 family, a reasoning language model for text and image input that the API docs describe as meant for cost-sensitive, high-volume work. Released on 9 July 2026 in the API, ChatGPT, Codex and Amazon Bedrock, it became the default ChatGPT model for Free and Go users in August 2026. [1] [2] [3] [4] [5] Primary source

Timeline of GPT-5.6 Luna →

Claims and evidence

  • Announced Announced with GPT-5.6 Sol and GPT-5.6 Terra as a limited preview for a small group of trusted partners. Primary source[6]first paragraph [7]Published June 26, 2026; 1. Introduction
  • Released Primary source[3]first paragraph; Availability and pricing [8]July, 2026: Jul 9 [1]Published July 9, 2026
  • Status AvailableListed in the API model catalog and not deprecated on the deprecations page as read on 2026-10-01. Primary source[2]Snapshots [9]Upcoming deprecations (gpt-5.6-luna appears only as a recommended replacement)
  • Successor of GPT-5.4 miniThe docs say GPT-5.6 Luna roughly corresponds to the nano model tier of earlier GPT-5 families and compare it with GPT-5.4 nano, an inline variant of the GPT-5.4 mini record; the Codex notice of 2026-07-31 names gpt-5.6-luna as the replacement for gpt-5.4-mini. Primary source[2]description; Quick comparison [10]2026-07-31: GPT-5.4 and GPT-5.4 mini retire from Codex on August 31
  • Change · Context length Context window of 1,050,000 tokens, up from 400,000 in GPT-5.4 nano, the model of the nano tier that Luna corresponds to.Compared with GPT-5.4 mini Primary source[2]Model details [11]Model details
  • Change · Training data Knowledge cutoff of 16 February 2026, against 31 August 2025 for GPT-5.4 nano.Compared with GPT-5.4 mini Primary source[2]Model details [11]Model details
  • Change · Tool use Gains the API features added with GPT-5.6, including Programmatic Tool Calling and multi-agent orchestration in beta.Compared with GPT-5.4 mini Primary source[8]July, 2026: Jul 9
  • Input text, image Primary source[2]Model details
  • Output text Primary source[2]Model details
  • Feature Reasoning mode Primary source[2]description (reasoning.effort); Model details (reasoning token support) [1]2. Model Data and Training
  • Feature Function calling Primary source[2]Supported features
  • Feature Computer use Primary source[2]Supported tools
  • Context window 1,050,000 tokensThe docs give 922,000 maximum input tokens and 128,000 maximum output tokens. Primary source[2]Model details
  • Parameters Not disclosedNot disclosed by OpenAI. No claim madeNo source
  • Access API, consumer app, cloud partner Primary source[3]Availability and pricing [8]July, 2026: Jul 9 [4]July 13–17, 2026: Use GPT-5.6 through Amazon Bedrock [5]Safety and availability

Lineage

Predecessors

Successors

Based on

Not derived from another model.

Variants and derived

None recorded.

Siblings

None recorded.

All ancestors

All descendants

Variants

No variants recorded in this record.

Related AI Radar coverage

All model releases from OpenAI on AI Radar →