Skip to content
AIMODELRACE
OpenAILanguage modelNew

GPT-6 Luna

The free-tier GPT-6: fast, friendly and cheap enough for anything high-volume.

Compare head-to-headRace it in 3D

Overall Intelligence

#25

of 56
10.0 pts behind #1

0

races led

0

weeks at #1

78.7

score

Released
Sep 22, 2026
4 wk ago
Context
400K
64K max output
Price
$0.2 / $1.6
per 1M in/out · $0.55 blended
Speed
232 tok/s
0.9s to first token
Modalities
text · image · audio · video
outputs text
Weights
Closed
API and apps

The 30-second version

The free-tier GPT-6: quick, friendly and cheap enough for anything high-volume. Luna is what most ChatGPT users talk to by default.

Key takeaways
  • 1Very low price ($0.20 / $1.60) with GPT-6-generation quality on everyday tasks.
  • 2Around 230 tokens per second.
  • 3Great for chat, summarisation and simple writing at scale.
  • 4Not a reasoning or agentic-coding specialist.

What you need to know

Purpose

Broad, affordable access to GPT-6 for consumers and high-volume API use.

What OpenAI optimised for

Cost and speed for mass-market use, keeping the GPT-6 conversational feel.

Unique strengths

  • Pleasant conversational style at a tiny price.
  • Fast responses.
  • Good multilingual coverage.
Insight
How it differs from rivals. Versus Gemini 3.5 Flash: friendlier writing, slower and weaker at multimodal input. Versus Haiku 5.5: better chat, weaker computer use.

Where it ranks

Current snapshot · illustrative scores · click any race to watch it

CodingHard ReasoningMathScienceWriting & LanguageMultimodalTool Use & AgentsLong ContextTrust & FactualitySpeed & Cost

GPT-6 LunaBest on each axis

Overall Intelligence over time

#1 frontierGPT-6 LunaV4 ProHaiku 5.5
55.060.065.070.075.080.085.090.02025Apr ’25Jul ’25Oct ’252026Apr ’26Jul ’26Oct ’26

Best use cases

01

Consumer chat

Everyday Q&A, brainstorming and light writing.

02

Summaries at scale

Digesting emails, reviews or transcripts cheaply.

03

First-line support

Answering common questions before escalation.

Strengths

  • Cost

    One of the cheapest frontier-family models.

  • Conversation

    Inherits much of GPT-6's style.

Weaknesses

  • Hard reasoning

    Far behind the top tiers.

  • Context

    400K window, smaller than most rivals.

When to choose it, and when not to

GPT-6 LunaoverHaiku 5.5

You need natural conversation more than tool or computer use.

Pick something else when…

  • Complex coding or agent work.
  • Long documents beyond 400K tokens.

Update history

The distilled version first; open any release for the deeper breakdown.