Overview
GPT-6 Luna is the light tier from the same launch, aimed at high-volume jobs with a clear goal: summarization, extraction and answering straightforward questions. It costs $0.10 per million input tokens and $0.50 per million output. OpenAI’s comparison is blunt, saying Luna at higher effort levels matches GPT-5.6 Sol on factual reliability at roughly a hundredth of the cost. Artificial Analysis found it emits about 51,000 output tokens per Intelligence Index task, up from roughly 41,000 for the previous generation, so the low price comes from unit pricing rather than token thrift.
Key Features
- Low unit price: At $0.10 per million input tokens it is the cheapest tier in the GPT-6 family, undercutting Xiaomi’s MiMo-V2.6-Flash on input price.
- Factual reliability: On OpenAI’s internal test built from real conversations where users flagged an error, Luna at higher effort levels matches GPT-5.6 Sol, though the set is deliberately chosen to be error-prone.
- Multiple entry points: Available in ChatGPT Work and Codex on Plus, Pro, Business, Enterprise and Edu accounts, with Free and Go users getting it in the desktop app.
- Shared engineering work: It benefits from the same caching and inference improvements as Sol, so one set of changes covers both tiers.
Use Cases
- Document summarization and information extraction at batch scale, returning structured fields
- Classification and routing, sending requests to the right model or to a human queue
- Support and Q&A scenarios where the goal is clear and the answer is short
- Data cleanup: high-volume, low-cost, repetitive work where moderate trade-offs are acceptable
Pros
- Going to $0.10 input changes the cost structure of high-frequency batch work
- Scores 66.6% on DeepSWE, which OpenAI describes as comparable to Opus 5 and Fable 5 at medium effort while costing 93% and 96% less per task
- Free and Go users can reach it in the desktop app, so the trial barrier is low
- Cached reads get the same 90% discount, so repeated prefixes in long pipelines stop being billed over and over
Pricing
The API charges $0.10 per million input tokens and $0.50 per million output, with cached input reads at a 90% discount. On the ChatGPT side, Plus, Pro, Business, Enterprise and Edu accounts get it in ChatGPT Work and Codex, while Free and Go users get it in the desktop app. Check the official page for current prices and quotas.
Summary
GPT-6 Luna pushes the light tier to $0.10 input and $0.50 output per million tokens, roughly half of GPT-5.6 Luna. What it buys is not a benchmark lead: the Artificial Analysis Intelligence Index stays at 37, level with the previous generation, while task cost drops from $0.18 to $0.07. It fits into pipelines for the high-frequency steps that never needed a flagship model.
Version History
- OpenAI ships GPT-6 Luna at $0.10/$0.50 per million tokens (2026-09-22): GPT-6 Luna arrived alongside Sol, positioned for high-volume jobs with a clear goal. It costs $0.10 per million input tokens and $0.50 per million output as permanent pricing, down sharply from $0.20 and $1.20 for GPT-5.6 Luna. It ships in ChatGPT Work and Codex, with Free and Go users getting it in the desktop app. Artificial Analysis measured an Intelligence Index of 37, level with the previous generation, with task cost falling from $0.18 to $0.07. Output runs about 51,000 tokens per benchmark task, so the savings come from unit pricing rather than token efficiency.