OpenAI
OpenAI says GPT-6 Sol and Luna cost half of GPT-5.6 promo
OpenAI released GPT-6 Sol and GPT-6 Luna in the API Sept. 22, 2026, priced 50% below GPT-5.6 promotional rates, with Luna at $0.10 per 1M input tokens.
OpenAI released two new reasoning models, GPT-6 Sol and GPT-6 Luna, and said their API prices are 50% lower than GPT-5.6 promotional pricing. Sol costs $2 per 1M input tokens and $10 per 1M output tokens. Luna costs $0.10 and $0.50.
The company released both models in the API Sept. 22, 2026, according to the OpenAI API changelog. The API IDs are gpt-6-sol and gpt-6-luna. Both accept text and image inputs, generate text, and run through the Responses and Chat Completions APIs.
Pricing
The changelog lists standard prices per 1M tokens for prompts up to 272K input tokens. Sol is $2 for input, $0.20 for cached input and $10 for output. Luna is $0.10, $0.01 and $0.50.
OpenAI’s Sol model page, read Sept. 28, 2026, lists the modifiers. Prompts over 272K input tokens are billed at 2x the input and cache rates and 1.5x the output rate for the full request. Batch and Flex processing cost 50% of standard, and Fast mode costs 2x. Cache writes are billed at 1.25x the uncached input rate, or $2.50 for Sol.
OpenAI’s pricing page, also read Sept. 28, lists GPT-5.6 Sol at $4 for input, $0.40 for cached input and $20 for output at short context. That makes Sol’s output price half of its predecessor’s. The page says GPT-5.6 Sol’s promotional pricing is available at least through Nov. 21, 2026. In its announcement, as quoted in a post on the OpenAI Developer Community forum Sept. 22, OpenAI also said Sol and Luna build on the advances behind GPT-6 Astra.
What the models are
Sol has a 1,050,000-token context window, 128,000 maximum output tokens and an April 20, 2026 knowledge cutoff, according to its model page. Reasoning effort can be set to none, low, medium (the default), high, xhigh or max. The page describes Sol as “built for complex coding and agentic workflows.”
TechCrunch reported Sept. 22 that OpenAI described Luna as suited to “high-volume tasks with a clear goal, like summarizing documents, extracting information, or answering quick questions.” TechCrunch also reported that OpenAI attributed the lower prices to improvements in caching and inference.
OpenAI said Sol makes about half as many mistakes as its predecessor, TechCrunch reported. The figure comes from OpenAI’s internal evaluations, which use de-identified real-world conversations in which users flagged mistakes. It is not an independent benchmark, and no outside benchmark results for either model were part of the announcement coverage reviewed here.
Sol and Luna are available in ChatGPT Work, in Codex on paid plans and in the API. Luna is also available to Free and Go users in the desktop app.
An image fix
On Sept. 25, 2026, the API changelog said OpenAI fixed a bug in image encoding that had degraded image understanding in Sol and Luna. The fix improves results on visual tasks in the API and Codex, including computer use. OpenAI recommends that anyone sending image inputs rerun evaluations and retry affected workflows.
Analysis
The pricing page puts numbers on the gap inside OpenAI’s own lineup. Sol’s input rate of $2 is one-fifth of GPT-6 Astra’s $10, and Astra’s output price of $50 is five times Sol’s $10, all at standard short-context rates as of Sept. 28.
The pricing page also lists a GPT-5.6 Sol promotional pricing window through at least Nov. 21, and the same page lists the GPT-6 prices.