Skip to main content

GPT-5.6 Luna

OpenAI

๐Ÿง  Reasoning๐ŸŒ Web Search

The fastest, most cost-efficient GPT-5.6 tier. Built for speed and high-volume simple tasks.

Released July 9, 2026

Pricing

Input tokens$1.00/M
Output tokens$6.00/M

Capacity

Context window1.1M tokens
Max output128K tokens

Capabilities

โœ“Reasoning & Planning
โœ“Web Search
โœ—Open Source

Best Scores

GDPval-AA v21591.8 pts
SWE-bench Verified93.0%
GPQA Diamond92.3%
Terminal-Bench 2.184.7%
SWE-bench Pro62.7%

Best For

๐Ÿ“summarization
๐ŸŒtranslation
๐Ÿ’กbrainstorming

Benchmark Scores

Specialized Skills

GDPval-AA v2

Real paid work from 44 different jobs, such as law, nursing, and software. Judges compare two answers side by side without knowing which model wrote them, and the winner gains rating points. A typical human expert scores 1000, so a higher number means the work was picked over a human more often.

1591.8 pts

Provider-reported; no independent run recorded yet.

Software Engineering

SWE-bench Verified

Hands the AI bugs from actual software projects and counts how many it fixes. Like a coding job interview, but with real work.

93.0%

Independently measured by Vals AI (mini-swe-agent).

Terminal-Bench 2.1

Puts the AI in front of a computer terminal and asks it to finish multi-step tasks on its own. Measures how good an "AI agent" it is.

84.7%

Provider-reported; an independent run by Vals AI (Terminus 2) lands at 79.03%.

SWE-bench Pro

The harder version of the coding test. Bigger codebases, trickier bugs. Scores drop for everyone, so the gaps between models become clearer.

62.7%

Provider-reported; no independent run recorded yet.

Knowledge

GPQA Diamond

PhD-level science questions written so you cannot just Google the answer. Tests whether the model can reason about hard science.

92.3%

Provider-reported; no independent run recorded yet.

Humanity's Last Exam

PhD-level questions across many subjects. Tests deep reasoning on the hardest questions humans can ask.

37.2%

Independently measured by Artificial Analysis.

Reasoning

Artificial Analysis Intelligence Index

A frequently refreshed overall score made from nine modern tests: real work, tool use, terminal tasks, science, hard questions, and long-context reasoning. It is a scorecard rather than a percent correct.

51.2 pts

Provider-reported; no independent run recorded yet.

Why Choose GPT-5.6 Luna?

Luna is OpenAI's speed and efficiency tier. At $1/$6, it competes on price with Haiku while maintaining access to OpenAI's full feature set and 1M+ token context. Perfect for high-volume, cost-sensitive applications.

How It Compares

vs Claude Haiku 4.5

Luna is slightly more expensive but has more context; Haiku is simpler.

vs GPT-5.6 Terra

Luna is 2.5x cheaper; Terra is more capable.

You Might Also Like