APEX-Agents-AA ↗35.84%
GPT-5.6 Luna (Max) · APEX-Agents-AA ↗
OpenAI / MODEL EVIDENCE
GPT-5.6 Luna
64 benchmarks · 112 observations · 67 sources · released 2026-07-09
Scores at a glance
64 benchmarks · large dot: GPT-5.6 Luna; small dots: closest models where reported
EnterpriseOps-Gym-AA ↗40.82%
GPT-5.6 Luna (Max) · EnterpriseOps-Gym-AA ↗
ITBench-AA ↗40.32%
GPT-5.6 Luna (Max) · ITBench-AA · root-cause score ↗
Terminal-Bench 2.1 ↗80.90%
GPT-5.6 Luna (Max) · Terminal-Bench 2.1 ↗
GPQA Diamond ↗91.67 %
GPT-5.6 Luna · GPQA Diamond · Accuracy ↗
2 more reported results
91.6% · GPT-5.6 Luna (max) · GPQA diamond · Best score (across scorers) · % · Epoch AI · GPQA diamond ↗
82.32% · GPT-5.6 Luna (low) · GPQA diamond · Best score (across scorers) · % · Epoch AI · GPQA diamond ↗
GPT-5.6 Luna (max) · FrontierMath-Tiers-1-3-v2-Private · Best score (across scorers) · % ↗
1 more reported result
41.4% · GPT-5.6 Luna (low) · FrontierMath-Tiers-1-3-v2-Private · Best score (across scorers) · % · Epoch AI · FrontierMath-Tiers-1-3-v2-Private ↗
GPT-5.6 Luna (max) · FrontierMath-Tier-4-v2-Private · Best score (across scorers) · % ↗
GPT-5.6 Luna (max) · OTIS Mock AIME 2024-2025 · Best score (across scorers) · % ↗
1 more reported result
66.67% · GPT-5.6 Luna (low) · OTIS Mock AIME 2024-2025 · Best score (across scorers) · % · Epoch AI · OTIS Mock AIME 2024-2025 ↗
GPT-5.6 Luna (max) · SimpleQA Verified · Best score (across scorers) · % ↗
GPT-5.6 Luna (max) · Chess Puzzles · Best score (across scorers) · % ↗
1 more reported result
17% · GPT-5.6 Luna (low) · Chess Puzzles · Best score (across scorers) · % · Epoch AI · Chess Puzzles ↗
GPT-5.6 Luna (max) · Mystery Game Puzzles · Best score (across scorers) · % ↗
2 more reported results
17% · GPT-5.6 Luna (low) · Mystery Game Puzzles · Best score (across scorers) · % · Epoch AI · Mystery Game Puzzles ↗
12% · GPT-5.6 Luna (medium) · Mystery Game Puzzles · Best score (across scorers) · % · Epoch AI · Mystery Game Puzzles ↗
ARC-AGI-1 ↗88%
GPT-5.6 Luna (max) · ARC-AGI · Score · % ↗
4 more reported results
87.67% · GPT-5.6 Luna (xhigh) · ARC-AGI · Score · % · Epoch AI · ARC-AGI ↗
76.5% · GPT-5.6 Luna (high) · ARC-AGI · Score · % · Epoch AI · ARC-AGI ↗
56.5% · GPT-5.6 Luna (medium) · ARC-AGI · Score · % · Epoch AI · ARC-AGI ↗
34.17% · GPT-5.6 Luna (low) · ARC-AGI · Score · % · Epoch AI · ARC-AGI ↗
BALROG ↗45.6%
GPT-5.6 Luna (max) · Balrog · Average progress · % ↗
SimpleBench ↗46.8%
GPT-5.6 Luna (unknown thinking) · SimpleBench · Score (AVG@5) · % ↗
1 more reported result
46.8% · GPT-5.6 Luna (xhigh) · SimpleBench · Score (AVG@5) · % · Epoch AI · SimpleBench ↗
WeirdML (v2) ↗60.86%
GPT-5.6 Luna (high) · WeirdML · Accuracy · % ↗
GPT-5.6 Luna (unknown thinking) · APEX-Agents · Pass@1 score · % ↗
ARC-AGI-2 ↗59.54%
GPT-5.6 Luna (max) · ARC-AGI-2 · Score · % ↗
4 more reported results
47.64% · GPT-5.6 Luna (xhigh) · ARC-AGI-2 · Score · % · Epoch AI · ARC-AGI-2 ↗
29.31% · GPT-5.6 Luna (high) · ARC-AGI-2 · Score · % · Epoch AI · ARC-AGI-2 ↗
7.36% · GPT-5.6 Luna (medium) · ARC-AGI-2 · Score · % · Epoch AI · ARC-AGI-2 ↗
5.14% · GPT-5.6 Luna (low) · ARC-AGI-2 · Score · % · Epoch AI · ARC-AGI-2 ↗
CursorBench 4.0 ↗35.9%
GPT-5.6 Luna (max) · CursorBench · Score · % ↗
4 more reported results
33% · GPT-5.6 Luna (xhigh) · CursorBench · Score · % · Epoch AI · CursorBench ↗
29.4% · GPT-5.6 Luna (high) · CursorBench · Score · % · Epoch AI · CursorBench ↗
22.2% · GPT-5.6 Luna (medium) · CursorBench · Score · % · Epoch AI · CursorBench ↗
16% · GPT-5.6 Luna (low) · CursorBench · Score · % · Epoch AI · CursorBench ↗
CritPt ↗20.57%
GPT-5.6 Luna (max) · CritPt · Accuracy · % ↗
4 more reported results
20.57% · GPT-5.6 Luna (xhigh) · CritPt · Accuracy · % · Epoch AI · CritPt ↗
16.57% · GPT-5.6 Luna (high) · CritPt · Accuracy · % · Epoch AI · CritPt ↗
4.86% · GPT-5.6 Luna (medium) · CritPt · Accuracy · % · Epoch AI · CritPt ↗
2.57% · GPT-5.6 Luna (low) · CritPt · Accuracy · % · Epoch AI · CritPt ↗
Surface Evolver Bench ↗61.88%
GPT-5.6 Luna (medium) · Surface Evolver Bench · Mean score · % ↗
DeepSWE v1.1 ↗67.19%
GPT-5.6 Luna (max) · DeepSWE · Pass@1 · % ↗
4 more reported results
56.86% · GPT-5.6 Luna (xhigh) · DeepSWE · Pass@1 · % · Epoch AI · DeepSWE ↗
44.25% · GPT-5.6 Luna (high) · DeepSWE · Pass@1 · % · Epoch AI · DeepSWE ↗
11.28% · GPT-5.6 Luna (medium) · DeepSWE · Pass@1 · % · Epoch AI · DeepSWE ↗
1.55% · GPT-5.6 Luna (low) · DeepSWE · Pass@1 · % · Epoch AI · DeepSWE ↗
ProofBench ↗60 %
GPT-5.6 Luna · ProofBench · Accuracy ↗
1 more reported result
60% · GPT-5.6 Luna (max) · ProofBench · Accuracy · % · Epoch AI · ProofBench ↗
FrontierCode 1.1 ↗39.81%
GPT-5.6 Luna (unknown thinking) · FrontierCode · Main score · % ↗
LMCA ↗48.54%
GPT-5.6 Luna (max) · LMCA · Score · % ↗
4 more reported results
48.54% · GPT-5.6 Luna (xhigh) · LMCA · Score · % · Epoch AI · LMCA ↗
47.38% · GPT-5.6 Luna (high) · LMCA · Score · % · Epoch AI · LMCA ↗
44.25% · GPT-5.6 Luna (medium) · LMCA · Score · % · Epoch AI · LMCA ↗
43.79% · GPT-5.6 Luna (low) · LMCA · Score · % · Epoch AI · LMCA ↗
DTBench ↗89.07%
GPT-5.6 Luna (xhigh) · DTBench · Accuracy · % ↗
4 more reported results
88.8% · GPT-5.6 Luna (max) · DTBench · Accuracy · % · Epoch AI · DTBench ↗
86.67% · GPT-5.6 Luna (high) · DTBench · Accuracy · % · Epoch AI · DTBench ↗
83.2% · GPT-5.6 Luna (medium) · DTBench · Accuracy · % · Epoch AI · DTBench ↗
80% · GPT-5.6 Luna (low) · DTBench · Accuracy · % · Epoch AI · DTBench ↗
GPT-5.6 Luna (max) · Furniture Assembly · Best score (across scorers) · % ↗
GDP.pdf ↗22.7%
GPT-5.6 Luna (unknown thinking) · GDP.pdf · GDP.pdf score · % ↗
1 more reported result
22.7% · GPT-5.6 Luna (medium) · GDP.pdf · GDP.pdf score · % · Epoch AI · GDP.pdf ↗
SciCode ↗53.59%
GPT-5.6 Luna (max) · SciCode · Score · % ↗
4 more reported results
51.62% · GPT-5.6 Luna (high) · SciCode · Score · % · Epoch AI · SciCode ↗
50.46% · GPT-5.6 Luna (xhigh) · SciCode · Score · % · Epoch AI · SciCode ↗
46.76% · GPT-5.6 Luna (medium) · SciCode · Score · % · Epoch AI · SciCode ↗
46.06% · GPT-5.6 Luna (low) · SciCode · Score · % · Epoch AI · SciCode ↗
Text Arena (Coding) ↗1518.87 Elo
GPT-5.6 Luna (xhigh) · WebDev Arena · Elo ↗
Blueprint-Bench 2 ↗22.56%
GPT-5.6 Luna (unknown thinking) · Blueprint-Bench 2 · Score · % ↗
WeirdML (v3) ↗9.34%
GPT-5.6 Luna (xhigh) · WeirdML v3 · Max performance · % ↗
Epoch Capabilities Index (ECI) ↗156.50 ECI
GPT-5.6 Luna · General ECI · 26 October 2026 fit ↗
Vals Science Index ↗25.3 %
GPT-5.6 Luna · Vals Science Index · Accuracy ↗
Vals Multimodal Index ↗69.06 %
GPT-5.6 Luna · Vals Multimodal Index · Accuracy ↗
Legal Research Bench ↗36.54 %
GPT-5.6 Luna · Legal Research Bench · All-Pass ↗
LegalBench ↗84.03 %
GPT-5.6 Luna · LegalBench · Accuracy ↗
GPT-5.6 Luna · Harvey's Legal Agent Benchmark v1 · Accuracy ↗
Tax Agent Bench ↗24.82 %
GPT-5.6 Luna · Tax Agent Bench · Accuracy ↗
Vals Finance Agent v2 ↗55.04 %
GPT-5.6 Luna · Finance Agent v2 · Accuracy ↗
GPT-5.6 Luna · Excel Modeling Benchmark v1 · Accuracy ↗
CorpFin v2 ↗64.22 %
GPT-5.6 Luna · CorpFin · Accuracy ↗
MortgageTax & TaxEval v2 ↗67.29 %
GPT-5.6 Luna · MortgageTax · Accuracy ↗
1 more reported result
76.17 % · GPT-5.6 Luna · TaxEval · Accuracy · Vals AI · TaxEval ↗
MedCode ↗42.39 %
GPT-5.6 Luna · MedCode · Accuracy ↗
MedScribe ↗84.39 %
GPT-5.6 Luna · MedScribe · Accuracy ↗
SAFE-Teen ↗23%
GPT-5.6 Luna · Urgent safety · share of applicable ratings marked fail ↗
2 more reported results
21% · GPT-5.6 Luna · Health advice · share of applicable ratings marked fail · Vals AI · SAFE-Teen ↗
35% · GPT-5.6 Luna · Relationship boundaries · share of applicable ratings marked fail · Vals AI · SAFE-Teen ↗
GPT-5.6 Luna (Mini-SWE-agent) · Terminal-Bench-Science · Score ↗
BioMysteryBench ↗61.48 %
GPT-5.6 Luna · BioMysteryBench · Accuracy ↗
MysteryMechanism ↗14.41 %
GPT-5.6 Luna · MysteryMechanism · Accuracy ↗
MMLU-Pro ↗86.04 %
GPT-5.6 Luna · MMLU Pro · Accuracy ↗
MMMU-Pro ↗85.03 %
GPT-5.6 Luna · MMMU Pro · Accuracy ↗
GPT-5.6 Luna · SAGE · Accuracy ↗
Vibe Code Bench ↗22.59 %
GPT-5.6 Luna · Vibe Code Bench · Vibe Code Bench 1–100 ↗
1 more reported result
77.06 % · GPT-5.6 Luna (OpenHands) · Vibe Code Bench · original task set · Vals AI · Vibe Code Bench ↗
GPT-5.6 Luna · ProgramBench · Accuracy ↗
Terminal-Bench 4.0 ↗11.62 %
GPT-5.6 Luna (Mini-SWE-agent) · Terminal-Bench 4.0 · Score ↗
SkillsBench ↗60.45 %
GPT-5.6 Luna (OpenHands) · SkillsBench · Score ↗
IOI v1 & IOI v2 ↗61.78 %
GPT-5.6 Luna · IOI · Accuracy ↗
Code Migration v1.1 ↗44.55 %
GPT-5.6 Luna · Code Migration v1 · Accuracy ↗
GPT-5.6 Luna (Mini-SWE-agent) · SWE-bench Verified · Score ↗
CyberBench v1.1 ↗73.63 %
GPT-5.6 Luna · CyberBench · Accuracy ↗
SRE-Bench ↗1.53 %
GPT-5.6 Luna · SRE Bench · Accuracy ↗
Public Benefits Bench v1.1 ↗61.16 %
GPT-5.6 Luna · Public Benefits Bench · Accuracy ↗
GPT-5.6-Luna (max) · Humanity's Sixth Sense · Score ↗
DistressBench ↗72.68%
gpt-5.6-luna · DistressBench · Score ↗
Vending-Bench 2 ↗4094.712 USD
GPT-5.6 Luna · End-of-year balance ↗
Closest models by score
Similar positions on shared benchmark leaderboards. Smaller gaps are closer.
GPT-5.6 Terra →OpenAI
29 of 40 shared benchmark comparisons
GPT-5.5 →OpenAI
27 of 40 shared benchmark comparisons
DeepSeek V4.1 Flash →DeepSeek
25 of 40 shared benchmark comparisons
How similarity is measured
Mean absolute gap in percentile positions on shared benchmark settings; lower is closer. At least 60% overlap is required, with a minimum of three when available. Independent evaluations are used when there are at least three; otherwise published report comparisons are used. Repeated model observations within one setting are averaged before ranking. Writing-style traits and composite indices are excluded. This describes similarity, not an overall quality ranking.
Sources
Every plotted observation links to its evaluation. Reported configurations, benchmark versions and task subsets can differ; similarity describes shared scores, not training lift.
- Blueprint-Bench 2 | Andon Labs ↗
- ARC Prize - Leaderboard ↗
- Arena leaderboard — WebDev evaluation ↗
- APEX-Agents-AA Benchmark | Artificial Analysis ↗
- EnterpriseOps-Gym-AA Benchmark | Artificial Analysis ↗
- ITBench-AA Benchmark | Artificial Analysis ↗
- SciCode Benchmark | Artificial Analysis ↗
- Terminal-Bench 2.1 Benchmark | Artificial Analysis ↗
- BALROG: Benchmarking Agentic LLM and VLM Reasoning On Games ↗
All 13 authors
Paglieri, Davide, Cupiał, Bartłomiej, Coward, Samuel, Piterbarg, Ulyana, Wolczyk, Maciej, Khan, Akbir, Pignatelli, Eduardo, Kuciński, Łukasz, Pinto, Lerrel, Fergus, Rob, Foerster, Jakob Nicolaus, Parker-Holder, Jack, Rocktäschel, Tim
- FrontierCode Leaderboard ↗
- DTBench — Conceptual Reasoning Index ↗
- LMCA — Conceptual Reasoning Index ↗
- Cursor · CursorBench ↗
- DeepSWE ↗
- APEX-Agents ↗
- ARC-AGI-2 ↗
- Chess Puzzles ↗
- CritPt v1 ↗
- FrontierMath Tier 4 (v2) ↗
- FrontierMath Tiers 1-3 (v2) ↗
- Furniture Assembly Benchmark ↗
- GPQA Diamond ↗
- Mystery Game Puzzles ↗
- OTIS Mock AIME 2024-2025 ↗
- SimpleQA Verified ↗
- SimpleBench ↗
- Vending Bench 2 — methodology and published results ↗
- AI benchmarks & capabilities ↗
- WeirdML ↗
- DistressBench — evaluation methodology and results ↗
- Humanity's Sixth Sense — evaluation methodology and results ↗
- AI Model Benchmarks — LM Council ↗
- GDP.pdf Benchmark | Surge AI ↗
- BioMysteryBench — evaluation methodology and results ↗
- Code Migration v1 — evaluation methodology and results ↗
- CorpFin — evaluation methodology and results ↗
- CyberBench — evaluation methodology and results ↗
- Excel Modeling Benchmark v1 — evaluation methodology and results ↗
- Finance Agent v2 — evaluation methodology and results ↗
- GPQA Diamond — evaluation methodology and results ↗
- Harvey's Legal Agent Benchmark v1 — evaluation methodology and results ↗
- IOI — evaluation methodology and results ↗
- LegalBench — evaluation methodology and results ↗
- Legal Research Bench — evaluation methodology and results ↗
- MedCode — evaluation methodology and results ↗
- MedScribe — evaluation methodology and results ↗
- SAFE-Teen: Teen Mental Health & Safety | Vals AI A conversation branches after the chatbot turns down a request — evaluation methodology and results ↗
- MMLU Pro — evaluation methodology and results ↗
- MMMU Pro — evaluation methodology and results ↗
- MortgageTax — evaluation methodology and results ↗
- MysteryMechanism — evaluation methodology and results ↗
- ProgramBench — evaluation methodology and results ↗
- ProofBench — evaluation methodology and results ↗
- Public Benefits Bench — evaluation methodology and results ↗
- SAGE — evaluation methodology and results ↗
- Vals Science Index — evaluation methodology and results ↗
- SkillsBench — evaluation methodology and results ↗
- SRE Bench — evaluation methodology and results ↗
- SWE-bench Verified — evaluation methodology and results ↗
- Tax Agent Bench — evaluation methodology and results ↗
- TaxEval — evaluation methodology and results ↗
- Terminal-Bench 4.0 — evaluation methodology and results ↗
- Terminal-Bench-Science — evaluation methodology and results ↗
- Vals Multimodal Index — evaluation methodology and results ↗
- Vibe Code Bench 1-100 — evaluation methodology and results ↗
- Vibe Code Bench — evaluation methodology and results ↗
- Surface Evolver Bench ↗
Names used in source tables
GPT-5.6 Luna · GPT-5.6 Luna (Max) · GPT-5.6 Luna (Mini-SWE-agent) · GPT-5.6 Luna (OpenHands) · GPT-5.6 Luna (high) · GPT-5.6 Luna (low) · GPT-5.6 Luna (max) · GPT-5.6 Luna (medium) · GPT-5.6 Luna (unknown thinking) · GPT-5.6 Luna (xhigh) · GPT-5.6-Luna (max) · gpt-5.6-luna