Head-to-head comparison

GPT-5.6 Terra vs. GPT-5.6 Luna

Compare published benchmark results from matching versions and cohorts, API pricing, and technical specifications. Any documented test setup differences remain visible.

shared result series: 209
distinct benchmarks: 29
latest retrieval: 2026-09-04

First modelSecond model

Specifications

General specifications

Release date, availability, context window, architecture, and published parameter counts at a glance. Undisclosed details remain marked as unavailable.

Specifications for GPT-5.6 Terra and GPT-5.6 Luna
Attribute	GPT-5.6 Terra	GPT-5.6 Luna
Provider	OpenAI	OpenAI
Released	Jun 26, 2026	Jun 26, 2026
Availability	Active	Active
Model type	Proprietary	Proprietary
Parameters	Not published	Not published
Architecture	Not published	Not published
Context window	1,050,000 tokens	1,050,000 tokens
Knowledge cutoff	February 16, 2026	February 16, 2026
Technical sources	Model, Context, Knowledge cutoff	Model, Context, Knowledge cutoff

Pricing

API pricing compared

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.

API pricing for GPT-5.6 Terra and GPT-5.6 Luna
Price type	GPT-5.6 Terra	GPT-5.6 Luna
API input	$2 ≤272K / $4 >272K	$0.2 ≤272K / $0.4 >272K
API output	$12 ≤272K / $18 >272K	$1.2 ≤272K / $1.8 >272K
Cache read	$0.2 ≤272K / $0.4 >272K	$0.02 ≤272K / $0.04 >272K
Cache write (5 min.)	$2.5 ≤272K / $5 >272K	$0.25 ≤272K / $0.5 >272K
Cache write (1 hr.)	Not listed	Not listed
Price verified	08/11/2026 Source	08/11/2026 Source

GPT-5.6 TerraGPT-5.6 Luna

API inputUSD per 1M tokens, base tier

GPT-5.6 Terra$2

GPT-5.6 Luna$0.2

API outputUSD per 1M tokens, base tier

GPT-5.6 Terra$12

GPT-5.6 Luna$1.2

Cache readUSD per 1M tokens, base tier

GPT-5.6 Terra$0.2

GPT-5.6 Luna$0.02

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 GPT-5.6 Terra and GPT-5.6 Luna
Model	Provider	Purchase route and tariff	Input per 1M tokens	Output per 1M tokens	Source
GPT-5.6 Terra	OpenAI	Direct from developerUp to 272,000 context tokens	$2	$12	Source08/11/2026
	OpenAI	Direct from developerAbove 272,000 context tokens	$4	$18	Source08/11/2026
	Amazon Bedrock	Via OpenRouteramazon-bedrock/us-east-1	$2.2	$13.2	Source09/06/2026
	Azure	Via OpenRouterazure	$2	$12	Source09/06/2026
	Azure	Via OpenRouterazure/eu	$2.2	$13.2	Source09/06/2026
	Azure	Via OpenRouterazure/us	$2.2	$13.2	Source09/06/2026
	OpenAI	Via OpenRouteropenai	$2	$12	Source09/06/2026
	OpenAI	Via OpenRouteropenai/fast	$4	$24	Source09/06/2026
	OpenAI	Via OpenRouteropenai/flex	Lowest listed price: $1	Lowest listed price: $6	Source09/06/2026
GPT-5.6 Luna	OpenAI	Direct from developerUp to 272,000 context tokens	$0.2	$1.2	Source08/11/2026
	OpenAI	Direct from developerAbove 272,000 context tokens	$0.4	$1.8	Source08/11/2026
	Amazon Bedrock	Via OpenRouteramazon-bedrock/us-east-1	$0.22	$1.32	Source09/06/2026
	Azure	Via OpenRouterazure	$0.2	$1.2	Source09/06/2026
	Azure	Via OpenRouterazure/eu	$0.22	$1.32	Source09/06/2026
	Azure	Via OpenRouterazure/us	$0.22	$1.32	Source09/06/2026
	OpenAI	Via OpenRouteropenai	$0.2	$1.2	Source09/06/2026
	OpenAI	Via OpenRouteropenai/fast	$0.4	$2.4	Source09/06/2026
	OpenAI	Via OpenRouteropenai/flex	Lowest listed price: $0.1	Lowest listed price: $0.6	Source09/06/2026

Performance

Shared benchmarks

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.

GPT-5.6 Terra

GPT-5.6 Luna

Coding: 97.7 / 81.5 (6 benchmark families)

Reasoning: 99.7 / 81.4 (3 benchmark families)

Mathematics: 100 / 90.2 (2 benchmark families)

Knowledge: 100 / 99 (2 benchmark families)

Finance: 99.6 / 99.7 (6 benchmark families)

Legal: 91.7 / 96.4 (4 benchmark families)

Agentic tasks: 98.5 / 74.2 (3 benchmark families)

Multimodal: 100 / 98.3 (1 benchmark family)

63 matched rows from 29 benchmark families were considered. The radar shows 8 of 14 comparable categories.

GPT-5.6 TerraGPT-5.6 Luna

ARC-AGI-2Overall. Higher is better. No documented setup difference

GPT-5.6 Terra67.1 %

GPT-5.6 Luna29.3 %

ARC-AGI-3RHAE overall score. Higher is better. Test setup not documented as identical

GPT-5.6 Terra0.49 %

GPT-5.6 Luna0.18 %

BrowseCompOverall. Higher is better. No documented setup difference

GPT-5.6 Terra87.5 %

GPT-5.6 Luna83.3 %

Code MigrationOverall. Higher is better. No documented setup difference

GPT-5.6 Terra47.804 %

GPT-5.6 Luna44.55 %

CorpFin v2Overall. Higher is better. No documented setup difference

GPT-5.6 Terra65.307 %

GPT-5.6 Luna64.219 %

EMBOverall. Higher is better. No documented setup difference

GPT-5.6 Terra66.205 %

GPT-5.6 Luna67.116 %

Finance Agent (v2)Overall. Higher is better. No documented setup difference

GPT-5.6 Terra54.436 %

GPT-5.6 Luna55.044 %

GPQA DiamondOverall. Higher is better. No documented setup difference

GPT-5.6 Terra90.909 %

GPT-5.6 Luna91.666 %

Harvey's Legal Agent BenchmarkOverall · Task fully resolved. Higher is better. No documented setup difference

GPT-5.6 Terra0.833 %

GPT-5.6 Luna1.25 %

IOIOverall. Higher is better. No documented setup difference

GPT-5.6 Terra65.25 %

GPT-5.6 Luna72.917 %

Legal Research BenchOverall · All-pass. Higher is better. No documented setup difference

GPT-5.6 Terra41.346 %

GPT-5.6 Luna36.538 %

LegalBenchOverall. Higher is better. No documented setup difference

GPT-5.6 Terra85.105 %

GPT-5.6 Luna84.034 %

Key to the marks:better valuebetter, but not conclusivetie

Benchmark scores for GPT-5.6 Terra and GPT-5.6 Luna
Benchmark and task	GPT-5.6 Terra	GPT-5.6 Luna	Source
ARC-AGI-2Overall Test details No documented difference in the test setupVersion: 2Metric: exact task accuracyScoring: Higher is betterSettings: Reasoning: high, Attempts: two exact output guesses per taskThe same verified test with five reasoning levels per model.	Winner67.1 %	29.3 %	ARC Prize Source details Data as of: 07/09/2026
ARC-AGI-3RHAE overall score Test details Test setup not documented as identicalNo direct winnerVersion: ARC-AGI-3 2026 verifiedMetric: RHAE scoreScoring: Higher is betterGPT-5.6 Terra: 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.	Higher measured value0.49 %	0.18 %	ARC Prize Source details Data as of: 08/26/2026
BrowseCompOverall Test details No documented difference in the test setupVersion: 1,266 tasksMetric: accuracyScoring: Higher is betterCross-vendor comparison table with browsing tools. Exact agent systems differ.	Winner87.5 % Editorial selection from the vendor table, not a complete extract of the comparison cohort.	83.3 % Editorial selection from the vendor table, not a complete extract of the comparison cohort.	OpenAI Source details Data as of: 07/09/2026
Code MigrationOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is better	Winner47.804 %	44.55 %	Vals AI Source details Data as of: 09/03/2026
CorpFin v2Overall Test details No documented difference in the test setupVersion: 2Metric: accuracyScoring: Higher is better	Winner65.307 %	64.219 %	Vals AI Source details Data as of: 08/12/2026
EMBOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is better	66.205 %	Winner67.116 %	Vals AI Source details Data as of: 09/03/2026
Finance Agent (v2)Overall Test details No documented difference in the test setupVersion: 2Metric: accuracyScoring: Higher is better	54.436 %	Winner55.044 %	Vals AI Source details Data as of: 09/03/2026
GPQA DiamondOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is better	90.909 %	Winner91.666 %	Vals AI Source details Data as of: 09/01/2026
Harvey's Legal Agent BenchmarkOverall · Task fully resolved Test details No documented difference in the test setupVersion: 1Metric: task resolution rateScoring: Higher is better	0.833 %	Winner1.25 %	Vals AI Source details Data as of: 09/03/2026
IOIOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is better	65.25 %	Winner72.917 %	Vals AI Source details Data as of: 08/09/2026
Legal Research BenchOverall · All-pass Test details No documented difference in the test setupVersion: 1Metric: all-pass rateScoring: Higher is better	Winner41.346 %	36.538 %	Vals AI Source details Data as of: 09/03/2026
LegalBenchOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is better	Winner85.105 %	84.034 %	Vals AI Source details Data as of: 09/01/2026
LMArena Text, Style ControlOverall Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-5.6 Terra: Snapshot: gpt-5.6-terra-xhigh, Reasoning: xhigh, Harness: Arena text, style control GPT-5.6 Luna: Snapshot: gpt-5.6-luna-xhigh, Reasoning: xhigh, Harness: Arena text, style controlGPT-5.6 Terra: Style-controlled Arena rating from 23,942 anonymous pairwise votes.GPT-5.6 Luna: Style-controlled Arena rating from 24,433 anonymous pairwise votes.	Winner1,466.233 95% confidence interval: ±5.062The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,452.364 95% confidence interval: ±5.03The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/01/2026
MedCodeOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is better	Winner43.414 %	42.391 %	Vals AI Source details Data as of: 09/03/2026
MedScribeOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is better	82.869 %	Winner84.391 %	Vals AI Source details Data as of: 09/03/2026
MMLU-ProOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is better	Winner86.659 %	86.036 %	Vals AI Source details Data as of: 09/01/2026
MMMU-ProOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is better	Winner86.474 %	85.029 %	Vals AI Source details Data as of: 09/01/2026
MortgageTaxOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is better	Winner67.329 %	67.29 %	Vals AI Source details Data as of: 09/01/2026
ProgramBenchOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is better	Winner0.5 %	0 %	Vals AI Source details Data as of: 09/01/2026
ProofBenchOverall Test details No documented difference in the test setupVersion: 1.1Metric: accuracyScoring: Higher is better	Winner74 %	60 %	Vals AI Source details Data as of: 09/01/2026
Public Benefits Bench v1.1Overall Test details No documented difference in the test setupVersion: 1.1Metric: accuracyScoring: Higher is better	Winner62.382 %	61.164 %	Vals AI Source details Data as of: 09/01/2026
SAGEOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is better	Winner47.003 %	44.22 %	Vals AI Source details Data as of: 09/03/2026
SkillsBenchOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Harness: OpenHands	58.895 %	Winner60.447 %	Vals AI Source details Data as of: 09/01/2026
SWE-bench VerifiedOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is better	Winner95.4 %	93 %	Vals AI Source details Data as of: 09/01/2026
TaxEval v2Overall Test details No documented difference in the test setupVersion: 2Metric: accuracyScoring: Higher is better	Tie76.166 %	Tie76.166 %	Vals AI Source details Data as of: 09/01/2026
Terminal-Bench 2.1Overall Test details No documented difference in the test setupVersion: 2.1Metric: accuracyScoring: Higher is better	77.528 %	Winner79.026 %	Vals AI Source details Data as of: 09/03/2026
Vals IndexOverall Test details No documented difference in the test setupVersion: 2Metric: weighted index scoreScoring: Higher is better	59.588 %	Winner59.881 %	Vals AI Source details Data as of: 09/03/2026
Vals Multimodal IndexOverall Test details No documented difference in the test setupVersion: 1.2Metric: weighted index scoreScoring: Higher is better	65.07 %	Winner69.058 %	Vals AI Source details Data as of: 08/11/2026
Vibe Code Bench v1.1Overall Test details No documented difference in the test setupVersion: 1.1Metric: accuracyScoring: Higher is betterSettings: Harness: OpenHands	74.594 %	Winner77.055 %	Vals AI Source details Data as of: 09/03/2026
ARC-AGI-2Overall Test details No documented difference in the test setupVersion: 2Metric: exact task accuracyScoring: Higher is betterSettings: Reasoning: low, Attempts: two exact output guesses per taskThe same verified test with five reasoning levels per model.	Winner18.8 %	5.1 %	ARC Prize Source details Data as of: 07/09/2026
ARC-AGI-3RHAE overall score Test details No documented difference in the test setupVersion: ARC-AGI-3 2026 verifiedMetric: RHAE scoreScoring: Higher is betterSettings: Reasoning: max, Agent: ARC Prize official harness, Harness: ARC Prize official harnessOfficial ARC Prize harness on unseen interactive environments. Reasoning levels remain separate.	Winner0.8 %	0.18 %	ARC Prize Source details Data as of: 08/26/2026
Code MigrationCLI Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	26.645 %	Higher measured value36.067 %	Vals AI Source details Data as of: 08/09/2026
CorpFin v2Exact Pages Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	63.17 %	Higher measured value63.52 %	Vals AI Source details Data as of: 08/06/2026
EMBComps Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	Higher measured value72.539 %	54.573 %	Vals AI Source details Data as of: 08/06/2026
Finance Agent (v2)Adjustments Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	50.442 %	Higher measured value52.027 %	Vals AI Source details Data as of: 08/06/2026
GPQA DiamondFew-Shot CoT Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	89.899 %	Higher measured value92.424 %	Vals AI Source details Data as of: 08/09/2026
Harvey's Legal Agent BenchmarkAntitrust Competition · Criteria passed Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: criteria pass rateScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	76.449 %	Higher measured value85.145 %	Vals AI Source details Data as of: 08/07/2026
IOIIOI 2024 Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	66.167 %	Higher measured value78 %	Vals AI Source details Data as of: 08/09/2026
Legal Research BenchAdministrative / Regulatory · All-pass Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: all-pass rateScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	Higher measured value57.576 %	48.485 %	Vals AI Source details Data as of: 08/06/2026
LegalBenchConclusion Tasks Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	Higher measured value90.327 %	88.864 %	Vals AI Source details Data as of: 08/09/2026
LMArena Text, Style ControlBusiness, management and finance Test details No documented difference in the test setupNo clear leadVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-5.6 Terra: Snapshot: gpt-5.6-terra-xhigh, Reasoning: xhigh, Harness: Arena text, style control GPT-5.6 Luna: Snapshot: gpt-5.6-luna-xhigh, Reasoning: xhigh, Harness: Arena text, style controlGPT-5.6 Terra: Style-controlled Arena rating from 4,559 anonymous pairwise votes.GPT-5.6 Luna: Style-controlled Arena rating from 4,691 anonymous pairwise votes.	No clear lead1,467.133 95% confidence interval: ±9.403The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,461.709 95% confidence interval: ±9.371The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/01/2026
MMLU-ProBiology Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	93.026 %	Higher measured value93.305 %	Vals AI Source details Data as of: 08/09/2026
MortgageTaxNumerical Extraction Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	74.324 %	Higher measured value75.596 %	Vals AI Source details Data as of: 08/09/2026
ProgramBenchAlmost Resolved Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	6.5 %	Higher measured value9 %	Vals AI Source details Data as of: 08/09/2026
SAGEAutomata Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 30,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 30,000 tokens	Higher measured value38.889 %	31.35 %	Vals AI Source details Data as of: 08/09/2026
SkillsBenchNo Skills Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Harness: OpenHands, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Harness: OpenHands, Output limit: 128,000 tokens	44.208 %	Higher measured value45.189 %	Vals AI Source details Data as of: 08/07/2026
SWE-bench Verified<15 min fix Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	77.835 %	Higher measured value96.392 %	Vals AI Source details Data as of: 08/08/2026
TaxEval v2Correctness Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	65.822 %	Higher measured value66.558 %	Vals AI Source details Data as of: 08/09/2026
Terminal-Bench 2.1Easy Test details Test setup not documented as identicalNo direct winnerVersion: 2.1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	No direct winner83.333 %	No direct winner83.333 %	Vals AI Source details Data as of: 08/06/2026
Vals IndexCorpFin v2 Test details Test setup not documented as identicalNo direct winnerVersion: 1.2Metric: benchmark scoreScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	Higher measured value66.434 %	65.035 %	Vals AI Source details Data as of: 08/09/2026
Vals Multimodal IndexCorpFin v2 Test details Test setup not documented as identicalNo direct winnerVersion: 1.2Metric: benchmark scoreScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	Higher measured value66.434 %	65.035 %	Vals AI Source details Data as of: 08/09/2026
ARC-AGI-2Overall Test details No documented difference in the test setupVersion: 2Metric: exact task accuracyScoring: Higher is betterSettings: Reasoning: max, Attempts: two exact output guesses per taskThe same verified test with five reasoning levels per model.	Winner83.9 %	59.5 %	ARC Prize Source details Data as of: 07/09/2026
ARC-AGI-3RHAE overall score Test details Test setup not documented as identicalNo direct winnerVersion: ARC-AGI-3 2026 verifiedMetric: RHAE scoreScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, 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.	Higher measured value0.65 %	0.18 %	ARC Prize Source details Data as of: 08/26/2026
Code MigrationCOBOL Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	65.844 %	Higher measured value70 %	Vals AI Source details Data as of: 08/09/2026
CorpFin v2Max Fitting Context Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	Higher measured value66.317 %	64.103 %	Vals AI Source details Data as of: 08/06/2026
EMBDataroom Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	81.591 %	Higher measured value83.009 %	Vals AI Source details Data as of: 08/06/2026
Finance Agent (v2)All-Pass Test details Test setup not documented as identicalNo direct winnerVersion: 2Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	40.448 %	Higher measured value43.285 %	Vals AI Source details Data as of: 08/06/2026
GPQA DiamondZero-Shot CoT Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	Higher measured value91.919 %	90.909 %	Vals AI Source details Data as of: 08/09/2026
Harvey's Legal Agent BenchmarkAntitrust Competition · Task fully resolved Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: task resolution rateScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	No direct winner0 %	No direct winner0 %	Vals AI Source details Data as of: 08/07/2026
IOIIOI 2025 Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	64.333 %	Higher measured value67.833 %	Vals AI Source details Data as of: 08/09/2026
Legal Research BenchAdministrative / Regulatory · Weighted partial credit Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: weighted partial-credit scoreScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	86.696 %	Higher measured value87.513 %	Vals AI Source details Data as of: 08/06/2026
LegalBenchInterpretation Tasks Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	Higher measured value83.178 %	82.142 %	Vals AI Source details Data as of: 08/09/2026
LMArena Text, Style ControlChinese Test details No documented difference in the test setupVersion: style-control-v1Metric: style-controlled Bradley-Terry ratingScoring: Higher is betterGPT-5.6 Terra: Snapshot: gpt-5.6-terra-xhigh, Reasoning: xhigh, Harness: Arena text, style control GPT-5.6 Luna: Snapshot: gpt-5.6-luna-xhigh, Reasoning: xhigh, Harness: Arena text, style controlGPT-5.6 Terra: Style-controlled Arena rating from 1,627 anonymous pairwise votes.GPT-5.6 Luna: Style-controlled Arena rating from 1,691 anonymous pairwise votes.	Winner1,518.544 95% confidence interval: ±15.369The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	1,474.185 95% confidence interval: ±15.484The Arena score measures human preference rather than a fixed capability test. The confidence interval and model configuration are part of the result.	LMArena Source details Data as of: 09/01/2026
MMLU-ProBusiness Test details Test setup not documented as identicalNo direct winnerVersion: 1Metric: accuracyScoring: Higher is betterGPT-5.6 Terra: Reasoning: xhigh, Output limit: 128,000 tokens GPT-5.6 Luna: Reasoning: max, Output limit: 128,000 tokens	90.875 %	Higher measured value91.001 %	Vals AI Source details Data as of: 08/09/2026
Key to the marks:better valuebetter, but not conclusivetie

How to read this comparison

Winning one benchmark is not an overall verdict

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

Compare GPT-5.6 Terra and GPT-5.6 Luna with other leading models using the same data and benchmark logic.

View the complete LLM comparison