Compare Models
GPT-5.6 Luna

Release Date: Jul 9, 2026

Developer OpenAI 🇺🇸
Context Window 1M
Max Output Tokens 128k
Token Costs (in/out) $0.20/1.20
Weights Private
Input Modalities

Accuracy

59.88 % ± 1.09

Cost / Test (Vals Index)

$ 0.620

Latency

23 min 16 s

Vals Index
BenchmarksAccuracyRankings
Vals Index

0.0%

±1.09
5/46
Code Migration

0.0%

±4.24
7/49
CyberBench

0.0%

±4.72
2/22
EMB

0.0%

±2.94
7/46
Finance Agent (v2)

0.0%

±0.31
8/49
Legal Research Bench

0.0%

±3.35
17/49
MedCode

0.0%

±2.27
36/85
MedScribe

0.0%

±2.58
21/84
MortgageTax

0.0%

±0.92
26/95
ProofBench v1.1

0.0%

±4.90
7/23
SAGE

0.0%

±3.33
34/76
Public Benefits Bench v1.1

0.0%

±1.27
16/30
TaxEval v2

0.0%

±0.84
7/140
Vibe Code Bench v1.1

0.0%

±3.07
10/84
Harvey's Legal Agent Benchmark

0.0%

±0.83
30/50
GPQA Diamond

0.0%

±1.74
16/133
IOI

0.0%

±5.08
4/62
LegalBench

0.0%

±0.42
36/137
MMLU Pro

0.0%

±0.35
46/133
MMMU

0.0%

±0.86
22/89
ProgramBench

0.0%

±0.00
17/39
SkillsBench

0.0%

±4.75
6/28
SWE-bench

0.0%

±1.14
7/83
Terminal-Bench 2.1

0.0%

±0.99
5/54
Proprietary BenchmarksAcademic BenchmarksIndustry Partners
Vals
Default Provider : OpenAI
Temperature: Default
Top P: Default
Top K: Default
Max Output Tokens: 128,000
Reasoning Effort: max