GPT-5.6 Luna: OpenAI's Low-Cost GPT-5.6 Model, Reasoning Effort, Rate Limits and Pricing

What GPT-5.6 Luna is

GPT-5.6 Luna is a large language model from OpenAI, released on 9 July 2026 as the low-cost tier of the GPT-5.6 family, alongside GPT-5.6 Sol and GPT-5.6 Terra. OpenAI designed it for cost-sensitive, high-volume workloads and says it roughly corresponds to the nano tier of earlier GPT-5 families. Its API ID is gpt-5.6-luna, which is also its only snapshot. It takes text and images as input and returns text. OpenAI notes that a rate-limit entry for an account's usage tier does not confirm access, and points users to its model availability guide if Luna is missing from the Playground or the model list. On LMArena's text leaderboard GPT-5.6 Luna has 1453 Elo (rank 78, xhigh effort).

Context, output and reasoning

Luna has the same limits as the larger GPT-5.6 models: a 1.05M-token context window, at most 922,000 input tokens and 128,000 output tokens per request, and a knowledge cutoff of 16 February 2026. It is a reasoning model, with reasoning effort set to none, low, medium, high, xhigh or max and medium as the default. It shares the features introduced with the GPT-5.6 family, including programmatic tool calling, explicit prompt caching controls, persisted reasoning and multi-agent orchestration in beta for the Responses API.

Pricing

Luna costs $0.20 per million input tokens, $0.02 per million cached input tokens and $1.20 per million output tokens. Those rates date from 30 July 2026, when OpenAI cut Luna's price by 80% and Terra's by 20%. Prompts longer than 272K input tokens are billed at twice the input rate and 1.5 times the output rate for the whole request, and cache writes cost 1.25 times the uncached input rate. OpenAI's comparison table lists GPT-5.6 Terra at $2 and $12 per million tokens and GPT-5.4 nano at $0.20 and $1.25. Since 5 August 2026 Fast mode, at twice the standard price, also accepts Luna's long-context prompts over 272K tokens.

Endpoints, tools and rate limits

Luna works through the Chat Completions, Responses and Batch endpoints, and supports streaming, structured outputs, function calling, file search, image input, web search and prompt caching. Through the Responses API it can use web search, file search, image generation, code interpreter, hosted shell, apply patch, skills, computer use, MCP and tool search. Its default rate limits are higher than Terra's: 5,000 requests and two million tokens per minute on the Build tier, 10,000 requests and ten million tokens on the Launch tier, and 30,000 requests and 180 million tokens per minute on the Grow tier, with a batch queue limit of 15 billion tokens. On Amazon Bedrock it is listed as openai.gpt-5.6-luna.

Where it sits in OpenAI's line-up

On 22 September 2026 OpenAI released GPT-6 Luna, which its changelog prices at $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 272K tokens. OpenAI's current guide describes GPT-6 Luna as the model for focused, high-volume tasks, and in October 2026 made its new Decisions API generally available with GPT-6 Luna. GPT-5.6 Luna remains listed in OpenAI's model catalogue with its own snapshot.

Official Resources