Model

GPT-6 Luna

CLOSED

OpenAI · GPT-6 family · released Sep 22, 2026

API id gpt-6-luna, released 2026-09-22. It is the low-cost GPT-6 tier at $0.10/$0.50 per Mtok (cached $0.01), 50% below GPT-5.6 Luna's $0.20/$1.20, with a 1.05M context. It is the ChatGPT Free/Go default since 2026-10-07 and powers the Decisions API beta (2026-10-06). It shares the 2026-09-25 image-encoding bug fix with Sol. OpenAI printed only one Luna number (DeepSWE). The other rows come from Anthropic's Claude Haiku 5.5 comparison table.

ReasoningCodingVisionFunction callingTool useAgentic

2529.5

Elo · rank #40

Parameters

Undisclosed

Active params

—

Context

1.05M tokens

Architecture

Proprietary model; architecture and parameter count undisclosed; 1.05M-token context, selectable reasoning effort (none to max)

License

Proprietary

Languages

—

API price (in/out)

$0.1 / $0.5

Modalities

text · vision

Benchmark results

Bar shows position within the tracked field; marker = field best

DeepSWECoding66.6%#16

best: Gemini 4 Argon · 77.9%

FrontierCodeCoding42.4%#12

best: Claude Opus 5.5 · 54.6%

Terminal-Bench 4.0Agents16.4%#18

best: Claude Sonnet 5.5 · 70.6%

Run it locally

Closed weights — available via API only. No local deployment.

Input / M tok

$0.1

Output / M tok

$0.5

GPT-6 family

Elo progression across releases

GPT-6 AstraSep 20263012.7 GPT-6 LunaSep 20262529.5 GPT-6 SolSep 20262848.0

Resources

DOCShttps://developers.openai.com/api/docs/models/gpt-6-luna APIAvailable via OpenAI API

API price $0.1/$0.5 · each benchmark row carries its own source badge (see methodology)