Back

OpenAI

GPT-6 Luna

ReasoningClosed weights
Intelligence
21.5
Rank 224 of 744 measured models
Output speed
119 t/s
Rank 132 of 403 measured models
Price/1M
$0.20
Rank 161 of 659 measured models
Context window
1M
Rank 84 of 1199 measured models
Cost per task
$0.0045
Rank 8 of 193 measured models

GPT-6 Luna (Low) scores 21.5 on the Artificial Analysis Intelligence Index, ranking 224th of 744 measured models. The closest models on this index are KAT Coder Pro V2 (21.7), A.X-K2 (21.5), and Qwen3.6 27B (21.4). At $0.20 per million tokens (3:1 input/output blended price), it is cheaper than KAT Coder Pro V2 ($0.525) and Qwen3.6 27B ($1.35). With an output speed of 119 t/s, it is faster than Qwen3.6 27B (55.6 t/s).

Artificial Analysis Indices
Low
Intelligence21.5
Rank 224 of 744 measured models
Reasoning levels
Each point is a reasoning level of this model: more reasoning often raises the score but costs more. Select a point to open that level.
Intelligence Indexmost efficient ↗Cost per task
Capability indexes
Performance in specific industries, measured by Artificial Analysis
Finance & Accounting24.3
Rank 138 of 218 measured models
Strategy & Ops22.8
Rank 136 of 218 measured models
Legal25.5
Rank 146 of 218 measured models
Engineering22.8
Rank 149 of 217 measured models
Economics30.1
Rank 155 of 225 measured models
Decision Index
GPT-6 Luna (Low) · Artificial Analysis
Decision Index55.2Rank 15 of 26 measured models
AA-Omniscience-Decision · accuracy66.0
AA-Omniscience-Decision · non-hallucination0.0
HLE-MultiChoice22.2
When2Call77.0
WildGuardTest88.5
Cost per 1M tasks$131.05
Average time per task1.5 s
Decision Index leaderboard
Standard Benchmarks
HLE20.3%
Rank 217 of 664 measured models
SciCode46.9%
Rank 133 of 231 measured models
AA-LCR74.0%
Rank 170 of 575 measured models
Terminal-Bench v40.0%
Rank 151 of 218 measured models
Models at a similar level
The closest models on the Intelligence Index, one configuration per family.

Preferences and analytics

Rybbit: visits, performance, clicks, copies, forms and session replay (masked inputs), with your permission. Refusing does not limit the site.

Privacy

Allow navigation analytics?

With your permission, Rybbit measures visits, performance and interactions (clicks, copies, forms), including session replay, to improve BenchSift. Input values are masked in replays. Refusing does not limit the site.

Your choice is stored for 180 days. Change it at any time using “Privacy preferences” in the footer. Withdrawal reloads the page.

No choice stored: Rybbit is disabled.

Privacy