GPT-6 Luna: price, context window and Codex use

What OpenAI says it's for

OpenAI's Codex model guide suggests Luna for specific, high-volume tasks when you know what a good result looks like, such as extraction, classification, transformation and structured summaries. It recommends GPT-6.1 Sol for complex coding and agentic work, and GPT-6 Astra for the hardest end-to-end work. Luna accepts text and image input and returns text.

Pricing details

• $0.10 per million input tokens, $0.01 cached input, $0.125 cache writes and $0.50 output, per the model page.
• Prompts over 272K input tokens are billed at 2x the input and cache rates and 1.5x the output rate for the whole request, which works out to $0.20 input, $0.02 cached input and $0.75 output per million tokens.
• Batch and Flex cost 50% of Standard, and Fast mode costs 2x. Regional processing adds 10% where available.
• Luna is currently the only model in OpenAI's Decisions API, in public beta since 6 October 2026. There, input costs $0.10 per million tokens and there's no output charge.

GPT-6 Luna in Codex

• GPT-5.5 retired from Codex for ChatGPT sign-ins on 14 October 2026 at 19:00 UTC. OpenAI's replacement on Free and Go is GPT-6 Luna in the desktop app, when available; on Plus, Pro, Business, Enterprise and Edu it's GPT-6.1 Sol. The OpenAI API and Codex sessions that use an API key aren't affected.
• OpenAI's Codex pricing lists GPT-6 Luna at Standard speed in the desktop app for Free ($0) and Go ($8 a month), subject to rollout. Plus ($20 a month) includes GPT-6.1 Sol and GPT-6 Luna.
• OpenAI estimates 350 to 3,000 local messages per five-hour window with Luna on Plus and Standard Business, against 15 to 160 for GPT-6.1 Sol and 5 to 45 for GPT-6 Astra. These are estimates, not fixed limits, and weekly limits may also apply. Pro currently has no five-hour limit.
• Credit rate: 2.5 credits per million input tokens, 0.25 per million cached input tokens and 12.5 per million output tokens.
• Select it with codex -m gpt-6-luna or /model in the CLI. OpenAI suggests starting Luna at High reasoning; it supports efforts up to Max, but not Ultra.
• In Enterprise and Edu, an administrator must enable Luna first.

API notes

• Reasoning effort supports none, low, medium (the default), high, xhigh and max. GPT-6 Sol also supports none; GPT-6.1 Sol doesn't.
• Use the Responses API for built-in tools and function calling. Chat Completions supports function calling only with reasoning effort set to none.
• Knowledge cutoff: 18 May 2026, later than GPT-6 Sol (20 April 2026) and GPT-6.1 Sol (30 April 2026).
• Fine-tuning isn't supported. EU data residency is available with Standard, Fast, Flex and Batch processing.

GPT-6 Luna vs Claude Haiku 5.5

For prompts up to 100,000 tokens, both cost $0.10 per million input tokens and $0.50 output, and cached input costs $0.01 on both. The difference is where long-prompt pricing starts. Claude Haiku 5.5 moves to $0.50 / $2.50 for the whole request once a prompt is over 100,000 tokens, per Anthropic's pricing page. Luna keeps $0.10 / $0.50 up to 272K input tokens, then charges 2x input and 1.5x output for the whole request. Luna's context window is 1,050,000 tokens and Haiku 5.5's is 1M; both allow 128K output tokens.

Independent benchmarks

Results for GPT-6 Luna from open datasets (Arena, Epoch AI) are on our benchmarks page.