readable AI benchmarks

OpenAI

GPT-5.6 Luna (Non-reasoning)

non-reasoning · closed weights · Jul 9, 2026

Quality Score32.2
Value Score29.1
Reliability37.5
Cache Discount90%

Model specification

Reasoning
non-reasoning
Input modalities
text, image
Output modalities
text
Context window
1M tokens
Weights
closed weights

Token prices USD per 1M tokens

Input
$0.200
Output
$1.20
Cached Input
$0.020
Cached Output
$0.250
Cost per task
$0.012

Capability

Intelligence
26.8
Coding
39.3
Agentic
22.2
Omniscience
-24.9
Correct
28.6%
Blended price
$0.174

Answer outcomes

Correct 28.6%Incorrect 53.5%Abstained 17.8%

Artificial Analysis benchmarks

GDPval-AA v2
29%
τ³-Banking
10%
Terminal-Bench v2.1
39%
SciCode
40%
Humanity’s Last Exam
7%
GPQA Diamond
65%
CritPt
0%
AA-Omniscience
38%
AA-LCR
39%

Similar models