Back

OpenAI

GPT-5.6 Luna

ReasoningClosed weights
Intelligence
25.0
Rank 174 of 744 measured models
Coding
50.7
Rank 121 of 298 measured models
Output speed
122 t/s
Rank 128 of 403 measured models
Price/1M
$0.45
Rank 256 of 659 measured models
Context window
1M
Rank 84 of 1199 measured models
Cost per task
$0.0156
Rank 18 of 193 measured models

GPT-5.6 Luna (Medium) scores 25.0 on the Artificial Analysis Intelligence Index, ranking 174th of 744 measured models. The closest models on this index are MiMo-V2-Omni-0327 (25.1), Inkling (Xhigh) (25.0), and GPT-5 Codex (High) (24.9). At $0.45 per million tokens (3:1 input/output blended price), it is cheaper than Inkling (Xhigh) ($1.76) and GPT-5 Codex (High) ($3.44). With an output speed of 122 t/s, it is slower than Inkling (Xhigh) (178 t/s) and GPT-5 Codex (High) (166 t/s).

Artificial Analysis Indices
Medium
Intelligence25.0
Rank 174 of 744 measured models
Coding50.7
Rank 121 of 298 measured models
Agentic23.9
Rank 86 of 189 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 & Accounting27.0
Rank 126 of 218 measured models
Strategy & Ops28.5
Rank 121 of 218 measured models
Legal27.7
Rank 129 of 218 measured models
Healthcare & Medical24.2
Rank 64 of 82 measured models
Engineering26.2
Rank 132 of 217 measured models
Economics32.0
Rank 142 of 225 measured models
Standard Benchmarks
GPQA Diamond85.9%
Rank 124 of 619 measured models
HLE25.8%
Rank 182 of 664 measured models
SciCode46.8%
Rank 134 of 231 measured models
AA-LCR75.0%
Rank 157 of 575 measured models
Terminal-Bench 2.153.2%
Rank 118 of 244 measured models
Terminal-Bench v40.5%
Rank 136 of 218 measured models
τ-Bench Banking17.7%
Rank 109 of 211 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