RankingEffort benchmarks

GPT-6 Luna · Max

Compare models

150 measured results · 7 effort settings · 66 benchmark/harness combinations

Model evidence and sources →

Change model

Model

Current configurationSources

Compare with

Choose a second model and its effort to see both profiles and benchmark differences below.

Compare withModel and configuration

Capability profile

One configuration, seven capabilities

Comparison score · 0–100

AGPT-6 Luna · Max4 of 7 capabilities supported

Gaps are unknown, not zero. Hollow points are preliminary.

Agentic
A53.4Supported
Hard reasoning
A34.4Supported
Coding
A53.3Supported
Human pref
Unknown
Knowledge
A22.6Supported
Multimodal
A39.7Preliminary
Long context
A52.9Preliminary

Each profile belongs to the named configuration. Capability scores use a shared panel for all axes, distinct from the overall ranking panel; they are not accuracy percentages or direct head-to-head wins.

Capability score evidence

GPT-6 Luna · Max: 35.4 overall comparison score. Meets the overall evidence requirements.

6 areas measured. The available task mix affects this estimate; missing areas remain unknown.

Which sources carry the weight?

Epoch AI · 15.6% direct comparison weight · independent of the model provider
Cognition · 15.4% direct comparison weight · independent of the model provider
LiveBench · 15.2% direct comparison weight · independent of the model provider
ARC Prize · 14.7% direct comparison weight · independent of the model provider
Vals AI · 14.3% direct comparison weight · independent of the model provider
Artificial Analysis · 13.7% direct comparison weight · independent of the model provider
Terminal-Bench · 11.0% direct comparison weight · independent of the model provider

Which benchmark families carry the weight?

FrontierCode · 15.4% direct comparison weight
Terminal-Bench · 13.6% direct comparison weight
ARC-AGI · 7.7% direct comparison weight
arc-agi-3 · 7.0% direct comparison weight
Aime · 3.1% direct comparison weight
Other families (29) · 53.2% direct comparison weight

These shares describe direct comparison weight, not fractions of the final score or confidence. Source balancing does not give every family equal weight or remove correlations between overlapping sources.

How sensitive is this score?

The audit removes one source at a time and refits against the same overall reference panel. These published results use the matching dataset audited 2026-10-20.

Both ranks use the same surviving configurations for each removal. The cohort starts with 444 configurations supported in the original audit, including separate effort settings. These are diagnostic positions, not the default leaderboard ranks. Eligibility is not rechecked after removal.

What changes when a source is removed?
Removed source	Rank before → after	Rank movement	Cohort remaining / lost support	Score after (change)
Artificial Analysis	91 → 92	Down 1	203 remaining / 241 lost	34.4 (-1.1 points)
ARC Prize	100 → 94	Up 6	444 remaining / 0 lost	41.4 (+6.0 points)
Datacurve	100 → 101	Down 1	444 remaining / 0 lost	30.3 (-5.2 points)
LiveCodeBench	100 → 100	No movement	444 remaining / 0 lost	35.4 (-0.0 points)
Scale	100 → 102	Down 2	444 remaining / 0 lost	37.4 (+2.0 points)
Terminal-Bench	100 → 94	Up 6	444 remaining / 0 lost	37.5 (+2.0 points)
Cognition	100 → 113	Down 13	444 remaining / 0 lost	31.5 (-3.9 points)
LiveBench	100 → 93	Up 7	443 remaining / 1 lost	37.1 (+1.7 points)
SWE-rebench	100 → 101	Down 1	444 remaining / 0 lost	35.0 (-0.4 points)
Epoch AI	99 → 97	Up 2	431 remaining / 13 lost	37.3 (+1.9 points)
Vals AI	100 → 103	Down 3	443 remaining / 1 lost	33.2 (-2.3 points)
Provider reports · Anthropic	100 → 100	No movement	444 remaining / 0 lost	35.4 (-0.0 points)
Provider reports · Moonshot AI	100 → 101	Down 1	444 remaining / 0 lost	35.0 (-0.4 points)
Provider reports · Meta	100 → 100	No movement	444 remaining / 0 lost	35.4 (-0.0 points)
Provider reports · Mistral	100 → 100	No movement	444 remaining / 0 lost	35.4 (+0.0 points)
Provider reports · OpenAI	100 → 100	No movement	444 remaining / 0 lost	35.7 (+0.3 points)
All provider reports	100 → 101	Down 1	444 remaining / 0 lost	35.2 (-0.2 points)

Lost support means a configuration no longer has a comparison path to the complete reference panel. Its rank is unavailable. A surviving path alone does not establish leaderboard eligibility.

Overlapping sources may remain correlated even with source balancing. Leave-one-source-out movement is a sensitivity diagnostic, not a spread of source-specific ranks or a confidence measure. Scoring methodology · Dated configuration audit

Every benchmark family

Removed evidence	Recomputed score	Change
deepswe	30.3	-5.2
frontiercode	31.5	-3.9
arc-agi-3	37.8	+2.3
terminal-bench	37.6	+2.2
epoch-game-puzzles	37.4	+2.0
arc-agi	37.2	+1.8
frontiermath	33.8	-1.6
livebench-instructions	36.8	+1.3
aime	34.1	-1.3
simpleqa	36.7	+1.3
mmmu-pro	36.7	+1.3
hle	36.5	+1.1
livebench-coding	34.6	-0.8
vals-excel-modeling	34.7	-0.8
livebench-data	34.7	-0.7
vibe-code	34.7	-0.7
vals-legal-research	36.1	+0.6
mmlu-pro	36.0	+0.6
vals-finance-agent	36.0	+0.5
gpqa	34.9	-0.5
harvey	35.0	-0.4
multi-swe-bench	35.0	-0.4
swe-rebench	35.0	-0.4
vals-tax-agent	35.8	+0.3
vals-code-migration	35.1	-0.3
livebench-reasoning	35.7	+0.3
swe-atlas-qna	35.2	-0.3
omniscience	35.7	+0.2
livebench-math	35.6	+0.2
ifbench	35.3	-0.1
automationbench	35.3	-0.1
enigma-eval	35.3	-0.1
analyst-agent	35.3	-0.1
swe-atlas-test-writing	35.3	-0.1
livebench-language	35.5	+0.1
tau-bench	35.3	-0.1
critpt	35.5	+0.1
apex-agents	35.5	+0.1
gdp-pdf	35.4	-0.1
terminal-bench-science	35.4	-0.1
aa-lcr	35.5	+0.1
itbench	35.4	-0.1
enterprise-ops	35.5	+0.0
briefcase	35.4	-0.0
swe-atlas-refactoring	35.4	-0.0
gdpval	35.4	-0.0
gmmlu	35.4	+0.0
swe-bench-pro	35.4	+0.0
livecodebench	35.4	-0.0
math500	35.4	+0.0

What evidence is missing from the fit?

Across all collected settings for this model: 192 observations, 129 contributing, 0 matched without graph weight and 63 excluded. Counts do not establish rank eligibility.

11 · Benchmark family or source protocol has not been reviewed
6 · Original SciCode grading defects are under review; retained as supporting evidence, excluded from aggregate and capability scores. Corrected SciCode-Verified requires a separate protocol review. Sources: https://arxiv.org/abs/2608.04975 and https://artificialanalysis.ai/evaluations/scicode
4 · Evaluator has not been reviewed for independent admission
25 · Provider-published lab self-report. Shown as a labeled claim and not admitted as a matched board comparison.
4 · Safety failure behavior is not a capability or intelligence score.
11 · Supporting evidence outside the reviewed capability core
2 · No reviewed effort for this catalog observation; any separately collected effort measurement is counted separately

Inspect every observation and exclusion →

Benchmark results

These are published benchmark scores. Supported effort estimates also appear on the Capability leaderboard. Results from different harnesses occupy different rows. Missing cells stay unknown; no scores are copied between effort levels.

Benchmark / source	Low	Medium	High	XHigh	Max	None	Non-reasoning
AA-AnalystAgentArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Share of tasks solved on all 5 of 5 attempts · 80 tasks, 5 attempts per task · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	Not reported	Not reported	Not reported	Not reported	43.75% Reported settings GPT-6 Luna (Max); Model source: https://artificialanalysis.ai/models/gpt-6-luna	Not reported	Not reported
AA-Briefcase EloArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details AA-Briefcase v1.1 is an agentic knowledge work benchmark developed by Artificial Analysis. AA-Briefcase Elo is a combined metric that aggregates rubric pass rate, analytical quality Elo and presentation Elo · Higher is better · Evaluation results measured independently by Artificial Analysis Unit: elo. Source collected 2026-10-18.	741.34 Elo Reported settings GPT-6 Luna (Low); Model source: https://artificialanalysis.ai/models/gpt-6-luna-low	1,098.75 Elo Reported settings GPT-6 Luna (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-luna-medium	1,186.42 Elo Reported settings GPT-6 Luna (High); Model source: https://artificialanalysis.ai/models/gpt-6-luna-high	1,236.97 Elo Reported settings GPT-6 Luna (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-luna-xhigh	1,334.83 Elo Reported settings GPT-6 Luna (Max); Model source: https://artificialanalysis.ai/models/gpt-6-luna	Not reported	903.18 Elo Reported settings GPT-6 Luna (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-6-luna-non-reasoning
AA-LCR v1.1Artificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	74% Reported settings GPT-6 Luna (Low); Model source: https://artificialanalysis.ai/models/gpt-6-luna-low	78.33% Reported settings GPT-6 Luna (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-luna-medium	79.33% Reported settings GPT-6 Luna (High); Model source: https://artificialanalysis.ai/models/gpt-6-luna-high	80% Reported settings GPT-6 Luna (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-luna-xhigh	83.33% Reported settings GPT-6 Luna (Max); Model source: https://artificialanalysis.ai/models/gpt-6-luna	Not reported	39.67% Reported settings GPT-6 Luna (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-6-luna-non-reasoning
AutomationBench-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Share of task objectives completed with no guardrail violations · Higher is better · Benchmark developed by Zapier · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	12.05% Reported settings GPT-6 Luna (Low); Model source: https://artificialanalysis.ai/models/gpt-6-luna-low	40.46% Reported settings GPT-6 Luna (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-luna-medium	47.82% Reported settings GPT-6 Luna (High); Model source: https://artificialanalysis.ai/models/gpt-6-luna-high	47.8% Reported settings GPT-6 Luna (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-luna-xhigh	53.17% Reported settings GPT-6 Luna (Max); Model source: https://artificialanalysis.ai/models/gpt-6-luna	Not reported	9.14% Reported settings GPT-6 Luna (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-6-luna-non-reasoning
CritPtArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Benchmark developed by Argonne and UIUC, with contributions from 60+ researchers globally. CritPt is under review by its authors following external feedback · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	2.57% Reported settings GPT-6 Luna (Low); Model source: https://artificialanalysis.ai/models/gpt-6-luna-low	10.57% Reported settings GPT-6 Luna (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-luna-medium	15.43% Reported settings GPT-6 Luna (High); Model source: https://artificialanalysis.ai/models/gpt-6-luna-high	17.43% Reported settings GPT-6 Luna (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-luna-xhigh	19.43% Reported settings GPT-6 Luna (Max); Model source: https://artificialanalysis.ai/models/gpt-6-luna	Not reported	1.14% Reported settings GPT-6 Luna (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-6-luna-non-reasoning
GDP.pdf: All-pass RateArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Share of attempts where every atomic criterion passed · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	9% Reported settings GPT-6 Luna (Low); Model source: https://artificialanalysis.ai/models/gpt-6-luna-low	13.8% Reported settings GPT-6 Luna (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-luna-medium	18.2% Reported settings GPT-6 Luna (High); Model source: https://artificialanalysis.ai/models/gpt-6-luna-high	18.6% Reported settings GPT-6 Luna (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-luna-xhigh	22.8% Reported settings GPT-6 Luna (Max); Model source: https://artificialanalysis.ai/models/gpt-6-luna	Not reported	5.8% Reported settings GPT-6 Luna (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-6-luna-non-reasoning
GDPval-AA v2.1 LeaderboardArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Elo rating for performance on real-world work tasks · Anchored to DeepSeek V4.1 Flash (max) at 1600 · Higher is better · Evaluation results measured independently by Artificial Analysis Unit: elo. Source collected 2026-10-18.	1,033.12 Elo Reported settings GPT-6 Luna (Low); Model source: https://artificialanalysis.ai/models/gpt-6-luna-low	1,262.27 Elo Reported settings GPT-6 Luna (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-luna-medium	1,339.93 Elo Reported settings GPT-6 Luna (High); Model source: https://artificialanalysis.ai/models/gpt-6-luna-high	1,363.87 Elo Reported settings GPT-6 Luna (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-luna-xhigh	1,431.86 Elo Reported settings GPT-6 Luna (Max); Model source: https://artificialanalysis.ai/models/gpt-6-luna	Not reported	1,045.45 Elo Reported settings GPT-6 Luna (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-6-luna-non-reasoning
Harvey LAB-AA v1.1: Hallucination-Gated All-Pass RateArtificial Analysis · Stirrup; Harvey private dataset v1.1.0; 120 tasks across 24 practice areas; up to 200 turns; internet disabled; GPT-6 Sol, Grok 4.7 and Claude Opus 5.5 rubric judges; GPT-6 Sol High two-pass hallucination check; hallucination-gated all-pass averaged across judges Benchmark details Average over tasks of the share of the judge panel finding every rubric criterion passed, with a task zeroed on any material hallucination · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	Not reported	Not reported	Not reported	Not reported	3.33% Reported settings GPT-6 Luna (Max); Model source: https://artificialanalysis.ai/models/gpt-6-luna	Not reported	Not reported
Humanity's Last ExamArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	20.25% Reported settings GPT-6 Luna (Low); Model source: https://artificialanalysis.ai/models/gpt-6-luna-low	28.27% Reported settings GPT-6 Luna (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-luna-medium	32.95% Reported settings GPT-6 Luna (High); Model source: https://artificialanalysis.ai/models/gpt-6-luna-high	34.34% Reported settings GPT-6 Luna (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-luna-xhigh	38.51% Reported settings GPT-6 Luna (Max); Model source: https://artificialanalysis.ai/models/gpt-6-luna	Not reported	8.57% Reported settings GPT-6 Luna (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-6-luna-non-reasoning
MLCR-AAArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Share of tasks judged accurate, complete, and concise · Expert + compound (hard) sets · Benchmark developed by Wisedocs · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	Not reported	Not reported	Not reported	Not reported	16.11% Reported settings GPT-6 Luna (Max); Model source: https://artificialanalysis.ai/models/gpt-6-luna	Not reported	Not reported
MMMU-ProArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	75.95% Reported settings GPT-6 Luna (Low); Model source: https://artificialanalysis.ai/models/gpt-6-luna-low	77.86% Reported settings GPT-6 Luna (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-luna-medium	77.51% Reported settings GPT-6 Luna (High); Model source: https://artificialanalysis.ai/models/gpt-6-luna-high	79.71% Reported settings GPT-6 Luna (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-luna-xhigh	79.65% Reported settings GPT-6 Luna (Max); Model source: https://artificialanalysis.ai/models/gpt-6-luna	Not reported	62.08% Reported settings GPT-6 Luna (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-6-luna-non-reasoning
AA-Omniscience IndexArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details AA-Omniscience Index (higher is better) measures knowledge reliability and hallucination. It rewards correct answers, penalizes hallucinations, and has no penalty for refusing to answer. Scores range from -100 to 100, where 0 means as many correct as incorrect answers, and negative scores mean more incorrect than correct. · Evaluation results measured independently by Artificial Analysis Unit: index. Source collected 2026-10-18.	-9.17 Reported settings GPT-6 Luna (Low); Model source: https://artificialanalysis.ai/models/gpt-6-luna-low	-5.05 Reported settings GPT-6 Luna (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-luna-medium	-5.5 Reported settings GPT-6 Luna (High); Model source: https://artificialanalysis.ai/models/gpt-6-luna-high	-1.83 Reported settings GPT-6 Luna (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-luna-xhigh	0.65 Reported settings GPT-6 Luna (Max); Model source: https://artificialanalysis.ai/models/gpt-6-luna	Not reported	-21.9 Reported settings GPT-6 Luna (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-6-luna-non-reasoning
AA-Omniscience AccuracyArtificial Analysis · 6000 questions; 1 attempt; correct-answer share; GPT-5.6 Luna medium grader Benchmark details AA-Omniscience Accuracy (higher is better) measures the proportion of correctly answered questions out of all questions, regardless of whether the model chooses to answer · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	40.77% Reported settings GPT-6 Luna (Low); Model source: https://artificialanalysis.ai/models/gpt-6-luna-low	43.13% Reported settings GPT-6 Luna (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-luna-medium	42.8% Reported settings GPT-6 Luna (High); Model source: https://artificialanalysis.ai/models/gpt-6-luna-high	44.17% Reported settings GPT-6 Luna (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-luna-xhigh	43.78% Reported settings GPT-6 Luna (Max); Model source: https://artificialanalysis.ai/models/gpt-6-luna	Not reported	31.85% Reported settings GPT-6 Luna (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-6-luna-non-reasoning
SciCodeArtificial Analysis · Artificial Analysis published evaluation; see source for benchmark-specific settings Benchmark details Benchmark curated by scientists across 16 disciplines. SciCode is under review following an independent audit of the dataset · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	46.88% Reported settings GPT-6 Luna (Low); Model source: https://artificialanalysis.ai/models/gpt-6-luna-low	50.93% Reported settings GPT-6 Luna (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-luna-medium	50.35% Reported settings GPT-6 Luna (High); Model source: https://artificialanalysis.ai/models/gpt-6-luna-high	51.74% Reported settings GPT-6 Luna (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-luna-xhigh	54.63% Reported settings GPT-6 Luna (Max); Model source: https://artificialanalysis.ai/models/gpt-6-luna	Not reported	43.06% Reported settings GPT-6 Luna (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-6-luna-non-reasoning
Terminal-Bench-Science 0.1Artificial Analysis · mini-swe-agent; Terminal-Bench-Science 0.1.0; 70 tasks; pass@1 averaged over 3 repeats Benchmark details Share of tasks passed (pass@1), averaged over three repeats · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	Not reported	Not reported	Not reported	Not reported	8.57% Reported settings GPT-6 Luna (Max); Model source: https://artificialanalysis.ai/models/gpt-6-luna	Not reported	Not reported
Terminal-Bench v4.0Artificial Analysis · mini-SWE-agent v2.4.6; 66 tasks; pass@1 averaged over 3 repeats; 500 steps; task timeout up to 8 hours Benchmark details Benchmark developed by the Laude Institute, Stanford, and open-source contributors · Independently benchmarked by Artificial Analysis · Evaluation results measured independently by Artificial Analysis Unit: percent. Source collected 2026-10-18.	0% Reported settings GPT-6 Luna (Low); Model source: https://artificialanalysis.ai/models/gpt-6-luna-low	2.53% Reported settings GPT-6 Luna (Medium); Model source: https://artificialanalysis.ai/models/gpt-6-luna-medium	4.55% Reported settings GPT-6 Luna (High); Model source: https://artificialanalysis.ai/models/gpt-6-luna-high	8.08% Reported settings GPT-6 Luna (Xhigh); Model source: https://artificialanalysis.ai/models/gpt-6-luna-xhigh	12.63% Reported settings GPT-6 Luna (Max); Model source: https://artificialanalysis.ai/models/gpt-6-luna	Not reported	1.52% Reported settings GPT-6 Luna (Non-reasoning); Model source: https://artificialanalysis.ai/models/gpt-6-luna-non-reasoning
ARC-AGI-2ARC Prize · v2 Semi-Private Benchmark details Original public board. Configuration and task coverage may differ; inspect source. Unit: percent. Source collected 2026-10-18.	4.58% Reported settings GPT-6 Luna (Low); Source model ID: openai-gpt-6-luna-low; 120 tasks	18.06% Reported settings GPT-6 Luna (Medium); Source model ID: openai-gpt-6-luna-medium; 120 tasks	31.39% Reported settings GPT-6 Luna (High); Source model ID: openai-gpt-6-luna-high; 120 tasks	41.94% Reported settings GPT-6 Luna (XHigh); Source model ID: openai-gpt-6-luna-xhigh; 120 tasks	59.31% Reported settings GPT-6 Luna (Max); Source model ID: openai-gpt-6-luna-max; 120 tasks	0% Reported settings GPT-6 Luna (None); Source model ID: openai-gpt-6-luna-none; 120 tasks	Not reported
ARC-AGI-3 · provider-adapterARC Prize · Provider Adapter Benchmark details ARC-AGI-3; Provider Adapter. Open each result for its exact effort, task subset, metric and protocol details. Unit: percent. Source collected 2026-10-18.	0.24% Reported settings ARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group openai-gpt-6-luna; variant provider-adapter-low; public-demo task scores are excluded	0.32% Reported settings ARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group openai-gpt-6-luna; variant provider-adapter-medium; public-demo task scores are excluded	0.38% Reported settings ARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group openai-gpt-6-luna; variant provider-adapter-high; public-demo task scores are excluded	0.51% Reported settings ARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group openai-gpt-6-luna; variant provider-adapter-xhigh; public-demo task scores are excluded	0.59% Reported settings ARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group openai-gpt-6-luna; variant provider-adapter-max; public-demo task scores are excluded	0.05% Reported settings ARC-AGI-3 Semi-Private verified summary; Provider Adapter harness; source group openai-gpt-6-luna; variant provider-adapter-none; public-demo task scores are excluded	Not reported
ARC-AGI-3 · standardARC Prize · Standard Benchmark details ARC-AGI-3; Standard. Open each result for its exact effort, task subset, metric and protocol details. Unit: percent. Source collected 2026-10-18.	0.03% Reported settings ARC-AGI-3 Semi-Private verified summary; Standard harness; source group openai-gpt-6-luna; variant low; public-demo task scores are excluded	0.19% Reported settings ARC-AGI-3 Semi-Private verified summary; Standard harness; source group openai-gpt-6-luna; variant medium; public-demo task scores are excluded	0.18% Reported settings ARC-AGI-3 Semi-Private verified summary; Standard harness; source group openai-gpt-6-luna; variant high; public-demo task scores are excluded	0.16% Reported settings ARC-AGI-3 Semi-Private verified summary; Standard harness; source group openai-gpt-6-luna; variant xhigh; public-demo task scores are excluded	0.1% Reported settings ARC-AGI-3 Semi-Private verified summary; Standard harness; source group openai-gpt-6-luna; variant max; public-demo task scores are excluded	0.03% Reported settings ARC-AGI-3 Semi-Private verified summary; Standard harness; source group openai-gpt-6-luna; variant none; public-demo task scores are excluded	Not reported
FrontierCode 1.1 · extended-codexCognition · codex Benchmark details FrontierCode 1.1; codex. Open each result for its exact effort, task subset, metric and protocol details. Unit: percent. Source collected 2026-10-18.	39.07% Reported settings FrontierCode 1.1; extended; 150 tasks; codex; source effort low; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.0013	50.15% Reported settings FrontierCode 1.1; extended; 150 tasks; codex; source effort medium; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0	51.3% Reported settings FrontierCode 1.1; extended; 150 tasks; codex; source effort high; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.0013	51.43% Reported settings FrontierCode 1.1; extended; 150 tasks; codex; source effort xhigh; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0	56.1% Reported settings FrontierCode 1.1; extended; 150 tasks; codex; source effort max; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0	Not reported	Not reported
FrontierCode 1.1 · main-codexCognition · codex Benchmark details FrontierCode 1.1; codex. Open each result for its exact effort, task subset, metric and protocol details. Unit: percent. Source collected 2026-10-18.	25.66% Reported settings FrontierCode 1.1; main; 100 tasks; codex; source effort low; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0	35.53% Reported settings FrontierCode 1.1; main; 100 tasks; codex; source effort medium; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0	37.26% Reported settings FrontierCode 1.1; main; 100 tasks; codex; source effort high; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0.002	37.1% Reported settings FrontierCode 1.1; main; 100 tasks; codex; source effort xhigh; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0	42.42% Reported settings FrontierCode 1.1; main; 100 tasks; codex; source effort max; weighted rubric score, not pass rate; unfair internet runs zeroed by evaluator; flag rate 0	Not reported	Not reported
Terminal-Bench 4.0 · codexTerminal-Bench · Codex Benchmark details Terminal-Bench 4.0; Codex. Open each result for its exact effort, task subset, metric and protocol details. Unit: percent. Source collected 2026-10-18.	Not reported	Not reported	Not reported	Not reported	16.4% Reported settings Terminal-Bench 4.0.0; 66 tasks; Codex; effort max; reported margin ±2.7 percentage points, confidence level unspecified; release date 2026-09-22; board checked 2026-10-07	Not reported	Not reported
LiveBench · AMPS Hard · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-10-18.	Not reported	Not reported	Not reported	Not reported	98% Reported settings LiveBench official task evaluation; livebench-2026-06-25-AMPS_Hard; gpt-6-luna-max; checkpoint=gpt-6-luna	Not reported	Not reported
LiveBench · code completion · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-10-18.	Not reported	Not reported	Not reported	Not reported	80.44% Reported settings LiveBench official task evaluation; livebench-2026-06-25-code_completion; gpt-6-luna-max; checkpoint=gpt-6-luna	Not reported	Not reported
LiveBench · code generation · 2026-06-25LiveBench · LiveBench official task evaluation Benchmark details LiveBench 2026-06-25 task snapshot. Task scores stay separate and related tasks share their reviewed family budget. Explicit checkpoint and effort labels only. Unit: percent. Source collected 2026-10-18.	Not reported	Not reported	Not reported	Not reported	77.47% Reported settings LiveBench official task evaluation; livebench-2026-06-25-code_generation; gpt-6-luna-max; checkpoint=gpt-6-luna	Not reported	Not reported

What these comparisons mean

Each source reports its own effort labels. Matching labels do not establish equal compute budgets or identical fallback behavior. Open Reported settings to inspect the exact model name, agent and task coverage. Multiple results at one level remain visible rather than selecting the best. Source-reported None or Non-reasoning labels are not treated as evidence that an API supports those settings.

These measurements use their own dated source collection. The leaderboard fits identified configurations jointly; it contains no pooled model ratings. Explicit thinking budgets are separate configurations. Read the methodology.

Download all effort benchmark results · 8,915 results across 259 models · Collected 2026-10-18

GPT-6 Luna · Max

Compare with

One configuration, seven capabilities

Agentic

Hard reasoning

Coding

Human pref

Knowledge

Multimodal

Long context

Capability score evidence

Benchmark results