Compare published benchmark results from matching versions and cohorts, API pricing, and technical specifications. Any documented test setup differences remain visible.
shared result series
240
distinct benchmarks
35
latest retrieval
2026-09-29
Change model selection
Specifications
General specifications
Release date, availability, context window, architecture, and published parameter counts at a glance. Undisclosed details remain marked as unavailable.
The table compares official standard API rates per 1M tokens, including published cache pricing. Batch, fast, priority, regional surcharges, tool calls, and cloud platform pricing are excluded.
Key to the marks:better valuebetter, but not conclusivetie
Pricing by provider
Direct developer pricing and provider endpoint tariffs routed through OpenRouter are listed separately. Prices are in USD per 1M tokens. Context-dependent direct tiers plus regional, Flex, Priority, and other OpenRouter tariffs remain individually identifiable.
Provider pricing for Claude Opus 5 and GPT-5.6 Luna
Only published results with the same benchmark version, task, metric, and comparison cohort are paired. Different reasoning levels, agents, harnesses, or output limits appear directly in the table.
Capability profile from matched benchmarks
Each axis averages directly matched benchmark families. A value of 100 means the stronger value within this pair, not a universal quality score.
Claude Opus 5
GPT-5.6 Luna
Coding: 100 / 60.4 (7 benchmark families)
Reasoning: 100 / 98.1 (1 benchmark family)
Mathematics: 100 / 60.6 (1 benchmark family)
Knowledge: 100 / 93.9 (1 benchmark family)
Finance: 99.8 / 91.2 (6 benchmark families)
Legal: 100 / 60.5 (3 benchmark families)
Agentic tasks: 100 / 93.9 (3 benchmark families)
Multimodal: 100 / 94.6 (1 benchmark family)
37 matched rows from 32 benchmark families were considered. The radar shows 8 of 14 comparable categories.
Claude Opus 5GPT-5.6 Luna
ARC-AGI-3RHAE overall score. Higher is better. Test setup not documented as identical
Claude Opus 5Higher measured value30.16 %
GPT-5.6 Luna0.18 %
BioMysteryBenchOverall. Higher is better. No documented setup difference
Claude Opus 5Winner79.259 %
GPT-5.6 Luna61.481 %
Code MigrationOverall. Higher is better. No documented setup difference
Claude Opus 5Winner57.473 %
GPT-5.6 Luna44.55 %
CorpFin v2Overall. Higher is better. No documented setup difference
Claude Opus 5Winner73.194 %
GPT-5.6 Luna64.219 %
CyberBench v1.1Overall. Higher is better. No documented setup difference
Claude Opus 565.357 %
GPT-5.6 LunaWinner73.631 %
EMBOverall. Higher is better. No documented setup difference
Claude Opus 5Winner73.556 %
GPT-5.6 Luna67.116 %
Finance Agent (v2)Overall. Higher is better. No documented setup difference
Claude Opus 5Winner58.633 %
GPT-5.6 Luna55.044 %
GPQA DiamondOverall. Higher is better. No documented setup difference
Claude Opus 5Winner93.434 %
GPT-5.6 Luna91.666 %
Harvey's Legal Agent BenchmarkOverall · Task fully resolved. Higher is better. No documented setup difference
Claude Opus 5Winner6.667 %
GPT-5.6 Luna1.25 %
IOIOverall. Higher is better. No documented setup difference
Claude Opus 5Winner84.333 %
GPT-5.6 Luna61.778 %
Legal Research BenchOverall · All-pass. Higher is better. No documented setup difference
Claude Opus 5Winner55.288 %
GPT-5.6 Luna36.538 %
LegalBenchOverall. Higher is better. No documented setup difference
Claude Opus 5Winner86.974 %
GPT-5.6 Luna84.034 %
Key to the marks:better valuebetter, but not conclusivetie
Benchmark scores for Claude Opus 5 and GPT-5.6 Luna
Benchmark and task
Claude Opus 5
GPT-5.6 Luna
Source
ARC-AGI-3RHAE overall scoreTest details
Test setup not documented as identicalNo direct winnerVersion: ARC-AGI-3 2026 verifiedMetric: RHAE scoreScoring: Higher is betterClaude Opus 5: Reasoning: high, Agent: ARC Prize official harness, Harness: ARC Prize official harness GPT-5.6 Luna: Reasoning: max, Agent: ARC Prize official harness, Harness: ARC Prize official harnessOfficial ARC Prize harness on unseen interactive environments. Reasoning levels remain separate.
Test setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterClaude Opus 5: Snapshot: claude-opus-5-high, Reasoning: high, Harness: Arena text, style control GPT-5.6 Luna: Snapshot: gpt-5.6-luna-xhigh, Reasoning: xhigh, Harness: Arena text, style controlClaude Opus 5: Style-controlled Arena rating from 55,063 anonymous pairwise votes.GPT-5.6 Luna: Style-controlled Arena rating from 35,823 anonymous pairwise votes.
Higher measured value1,491.077
95% confidence interval: ±3.89The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.
1,453.977
95% confidence interval: ±4.487The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.
No documented difference in the test setupVersion: arxiv:2609.25804v2Metric: both-orders-correct, engineering/research macro averageScoring: Higher is betterSettings: Reasoning: default, Output limit: 65,536 tokens, Attempts: 502 questions, two option orders eachSame prompt for all models. Each of 502 questions is shown twice with the options reversed. Average of engineering and research success rates, requiring both orders to be correct. 65,536-token output budget; unparseable answers count as wrong.
Winner55.5 %
0 of 1,004 responses were unparseable and counted as wrong. Not a Gradually test.
49 %
0 of 1,004 responses were unparseable and counted as wrong. Not a Gradually test.
Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 1Metric: criteria pass rateScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Legal Research BenchAdministrative / Regulatory · All-passTest details
Test setup not documented as identicalNo direct winnerVersion: 1Metric: all-pass rateScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterClaude Opus 5: Snapshot: claude-opus-5-max, Reasoning: max, Harness: Arena text, style control GPT-5.6 Luna: Snapshot: gpt-5.6-luna-xhigh, Reasoning: xhigh, Harness: Arena text, style controlClaude Opus 5: Style-controlled Arena rating from 26,760 anonymous pairwise votes.GPT-5.6 Luna: Style-controlled Arena rating from 35,823 anonymous pairwise votes.
Higher measured value1,488.172
95% confidence interval: ±5.082The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.
1,453.977
95% confidence interval: ±4.487The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.
Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterClaude Opus 5: Temperature: 1, Output limit: 30,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 30,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 1.2Metric: benchmark scoreScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 1.2Metric: benchmark scoreScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Test setup not documented as identicalNo direct winnerVersion: 1Metric: task resolution rateScoring: Higher is betterGPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens
Key to the marks:better valuebetter, but not conclusivetie
Our model tests
Our tests
Compare how the models respond to the same prompt. Each test shows the first attempt, with no subsequent fixes to the generated code. These results do not contribute to an overall score.
Pelican on a bicycle
A community classic for free-form SVG drawing.
Claude Opus 5
The API response was not fully received. The model was not evaluated.
Reasoning (requested): high
First attempt
Run date
7 Sept 2026
API provider
openrouter
Requested API model
anthropic/claude-opus-5
Reasoning setting
high
Pinned endpoint
Anthropic | anthropic/claude-opus-5-20260723
Temperature
Provider default
Token limit including reasoning
8,192 tokens
Test protocol
community-visual-tests-v1
Technical checks
Failed
Not yet visually reviewed. A successful recording does not confirm correct physics or full compliance with the prompt.
GPT-5.6 Luna
A community classic for free-form SVG drawing.
Reasoning (requested): high
First attempt
Run date
7 Sept 2026
API provider
openrouter
Requested API model
openai/gpt-5.6-luna
API model reported in response
openai/gpt-5.6-luna
Reasoning setting
high
Pinned endpoint
OpenAI | openai/gpt-5.6-luna-20260709
Temperature
Provider default
Token limit including reasoning
8,192 tokens
Test protocol
community-visual-tests-v1
Technical checks
Passed, not a quality rating
Response time
49.9 s
Visually reviewed. This individual test does not establish overall model quality.
Your workload, budget, and required context length matter most. A coding benchmark says little about visual understanding or agent performance.
The standard price table uses direct developer rates. The provider comparison labels OpenRouter endpoints, regions, and special tariffs separately.
Every result links to its measurement source and retrieval date. Labels distinguish vendor reports, official benchmark leaderboards, and independent evaluations. Disputed or archived results are excluded.
More matchups
Related LLM comparisons
Compare Claude Opus 5 and GPT-5.6 Luna with other leading models using the same data and benchmark logic.
