Atlas

Compare Models

Opus 4.8 vs GPT-5.6 Terra

GPT-5.6 Terra leads by 1.0 capability points (≈95% interval −1.0 to +2.9): statistically tied · P(higher) ≈ 83%

Opus 4.8 wins 127 of 260 shared benchmarks (110 lost, 23 tied)

Capability vs. relative cost
Opus 4.8GPT-5.6 TerraClosed weightsOpen weights

Each line runs through one model’s priced reasoning settings, cheapest to most expensive; grey points are other models’ priced settings in the same range. Results whose reasoning effort the source didn’t report are left out wherever a model has named effort levels. Capability and relative cost are the published estimates (relative cost is cost per task as a multiple of GPT-5 (high)); hover a setting, or use the arrow keys, to set it beside the other model’s same setting.

Opus 4.8vsGPT-5.6 Terracapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins127 – 110 · 260 shared, 23 tied

Comparing matching setups only (84 benchmarks excluded) · suites counted once
← Opus 4.8GPT-5.6 Terra →

Opus 4.8vsGPT-5.6 Terrawins by categoryData Nov 15, 2026Data: Vals AI, Mercor, Artificial Analysis and 69 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-5.6 Terra minus Opus 4.8.

HPCT Refusal (CAIS)Safety85.025.0−60.0
BullshitBench v2 LeaderboardSafety94.054.0−40.0
Harvey's Legal Agent Benchmark - Healthcare Life SciencesLegal40.00.0−40.0
YC-BenchGames1711389636144−37.5
LLM Chess EloGames4501293+34.6
Concept Synth INDUCTION Challenge64 Round 1General QA6.337.5+31.3
VCT-Refusal (CAIS)Safety41.210.7−30.5
PMPP-HardCode62.331.9−30.4
Gray Swan IPI Attack Success (k=15)Safety8.037.3−29.3
BioTIER RefusalSafety94.765.6−29.1
Aikido Known-CVE Detection (Pass@3 Union)Cybersecurity61.588.5+26.9
Opus Magnum CampaignGames22.847.7+24.9
HardGeoBench Avg DistanceMultimodal12451991−23.0
PointerbenchMultimodal68.947.3−21.6
HLE Extended (Mercor held-out)General QA49.470.7+21.3
SnakeBench Best ApplesGames28.021.0−20.6
LMArena Agent Arena Net ImprovementAgents6.90.6−20.5
Harvey's Legal Agent Benchmark - Capital MarketsLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Data Privacy CybersecurityLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Emerging Companies Venture CapitalLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Energy Natural ResourcesLegal20.00.0−20.0
RiemannBenchMath47.266.4+19.2
LisanBench Path LengthGames94633303−19.1
ReactBench v1Code36.553.3+16.9
ChessBench — Historical Output RepairGames24.040.7+16.7
Visual Acuity 4K Threshold RowMultimodal10.012.0+16.7
Blueprint-Bench 2Multimodal0.10.3+16.3
LisanBench Difficulty-WeightedGames26941197−16.0
MMMU-Pro Extended (Mercor held-out)Multimodal31.447.4+16.0
SWE-bench Verified Extended (Mercor held-out)Code63.378.7+15.4
SnakeBench RatingGames35.429.8−14.9
TxBench Antibody Discovery V0Science45.631.0−14.7
FrontierMath Tier 4 (v2) - PrivateMath56.170.7+14.6
AA-Omniscience IndexIndexes28.80.1−14.3
SM BenchSafety72.959.0−13.9
Humanity's Last ExamGeneral QA49.835.9−13.9
Medical Long Context Reasoning (MLCR-AA)Health & Medicine45.631.7−13.9
SpeciEvalSafety84.398.1+13.8
SpeciEval (Refusal-Aware Retries, Historical)Safety84.398.1+13.8
SpeciEval (Pooled English Runs)Safety84.398.1+13.7
Terminal-Bench 4.0 (Artificial Analysis)Code21.735.4+13.6
Toloka Arena (v1)Agents54.341.4−12.9
APEX-SWECode43.931.1−12.8
EQ-Bench 4Chat & Writing12681220−12.5
DeepsecBenchCode6.919.2+12.3
DeepsecBench — Revalidated V1Code6.919.1+12.2
DeepsecBench — Revalidated V1 (233 issues)Code6.919.1+12.2
BrowseComp (Mercor 130-question subset)Search73.785.8+12.1
CRED v0.92Science85.097.0+12.0
SnakeBench Average ApplesGames15.112.7−12.0
ARC-AGI-2General QA72.183.9+11.8
GDPval Extended (Mercor held-out rubric score)Professional Work58.947.2−11.7
Furniture Assembly BenchmarkMultimodal42.554.2+11.7
AA-LCR Extended (Mercor held-out)Search63.274.4+11.2
BioSecBench-Function V0Science44.633.5−11.1
Machiavelli Deception (CAIS)Safety83.780.4+10.2
ChessBench CoherenceGames79.769.7−10.0
SnakeBench Win RateGames60.050.0−10.0
MedCodeHealth & Medicine53.243.4−9.8
Extended NYT Connections 940-Puzzle SetGames94.184.7−9.4
NoCoT-Bench NCKI (kspine v3)General QA117104−9.3
GBENCH One-shot Coding GScoreGames54.063.3+9.3
ScBench-Long V0 VerifiedScience25.834.9+9.1
SpatialBench-Long V0 VerifiedScience28.837.9+9.1
GraphWalks BFS - 256KGeneral QA85.976.9−9.0
GameBench 2Games62.553.5−9.0
IFBench (Artificial Analysis avg@5)Chat & Writing62.271.2+9.0
HardGeoBench Country AccuracyMultimodal53.845.0−8.7
BuseyBench SVGMultimodal5.26.1+8.7
A-Fantasia (Valid Attempts with Retries)Games34.743.3−8.7
AHK-EvalCode80.688.9+8.3
τ²-bench (Telecom)Agents94.486.3−8.2
Vibe Code Bench v1.1Code82.774.6−8.1
Harvey HLAB — Criteria Pass RateLegal87.979.8−8.1
RedlineBenchLegal39.747.8+8.1
Roboflow Vision EvalsMultimodal68.176.2+8.1
BabaIsBench Stage 0: The IntroGames96.088.0−8.0
SWE Atlas Codebase QnA Extended (Mercor held-out)Code24.016.0−8.0
SAGEMath54.847.0−7.8
APEX-AccountingFinance & Accounting48.040.3−7.7
Olam Social Poker EloGames15181511−7.6
PCT Political Inconsistency (CAIS)Safety53.846.2+7.6
Agent Red Teaming k=100 (CAIS)Safety47.054.6−7.6
SnakeBench Total ApplesGames226127−7.5
APEX-Agents 1.1Professional Work64.471.8+7.4
GeneBench-ProScience16.023.3+7.3
NoCoT-Bench NCRI 15.2General QA10289.0−7.2
WCB (SecureBio)Science66.759.8−6.9
SWE-bench VerifiedCode88.695.4+6.8
APEX-1 (Mercor, mean score)Professional Work62.769.5+6.8
A-FantasiaGeneral QA50.043.3+6.7
Finance Agent v2 - Market AnalysisFinance & Accounting67.260.6−6.6
HieroglyphBenchMultimodal14.020.5+6.5
Bio Capabilities Index (SecureBio)Indexes164155−6.5
IntPhys 2Multimodal55.161.3+6.2
Document ArenaProfessional Work14751468−6.2
TxBench Oligo Discovery V0Science39.433.3−6.1
𝜏³-BankingAgents34.240.2+6.0
FrontierMath Tiers 1-3 (v2) - PrivateMath80.086.0+6.0
Vals Multimodal IndexIndexes70.965.1−5.8
DeepSWE v1.1Code54.460.2+5.8
Public Benefits Bench v1.1Professional Work68.162.4−5.8
Humanity's Last Exam DiamondGeneral QA23.128.8+5.7
BioSecBench-RefusalSafety28.634.3+5.7
Terminal-Bench 2.1Code71.977.5+5.6
EnterpriseOps-Gym-AAAgents44.038.5−5.5
CharXiv ReasoningMultimodal80.585.9+5.4
Vals IndexIndexes70.465.1−5.2
Finance Agent v2 - Earnings AnalysisFinance & Accounting72.177.3+5.2
SkillsBench without SkillsAgents43.748.8+5.2
MMMU ProMultimodal75.680.7+5.1
HPCT (SecureBio)Science63.168.2+5.1
MRCR v2 (8-needle, Context Arena) - 128K pointwiseGeneral QA75.280.2+5.0
Vision Arena OverallMultimodal12861266−5.0
DistressBenchSafety71.476.4+5.0
Chess Puzzles (task version 2.0)Games52.047.0−5.0
HealthBench Professional (Length-Adjusted)Health & Medicine53.057.7+4.7
BrowseComp Extended (Mercor held-out)Search30.025.3−4.7
WeirdML (v2)Science82.978.3−4.6
SUPERChemScience60.364.9+4.6
OSWorld 2.0 500-step Partial ScoreAgents54.850.2−4.6
PubMedQA (SecureBio)Health & Medicine75.980.4+4.5
AdvancedIFChat & Writing73.778.1+4.4
CharXiv Extended (Mercor held-out)Multimodal76.781.1+4.4
GPQA-Bio (SecureBio)Science80.985.3+4.4
TextQuests Average Cumulative HarmSafety21.022.2−4.2
ERQAMultimodal59.563.6+4.1
SpatialBench V0 VerifiedScience66.470.4+4.1
ARC-AGI-1 Semi-PrivateGeneral QA92.596.5+4.0
VCT v2 (SecureBio)Science45.941.9−4.0
GDPval (Mercor 213-task rubric score)Professional Work84.580.7−3.8
Surface Evolver BenchAgents87.583.8−3.7
SimpleQA VerifiedGeneral QA39.543.2+3.7
Terminal-Bench 2.1 Extended (Mercor held-out)Code32.736.4+3.7
Conceptual Reasoning IndexIndexes68.365.7−3.6
CorpFin v2 - Max Fitting ContextFinance & Accounting69.866.3−3.5
LAB-Bench CloningScenarios (SecureBio)Science95.992.5−3.4
VariantBenchScience42.138.7−3.4
Agents' Last Exam Overall ScoreAgents45.248.5+3.3
MedXpertQA MM (200-question set)Health & Medicine72.575.7+3.2
ScBench-LongScience20.623.8+3.2
LMArena Agent Arena Bash RecoveryAgents7.20.9−3.2
Excel Modeling BenchmarkFinance & Accounting69.466.2−3.2
GraphWalks BFS - 1MGeneral QA68.171.2+3.1
Senior SWE-BenchCode30.527.4−3.1
WordleBenchGames94.091.0−3.0
MedScribeHealth & Medicine85.882.9−2.9
LMArena Agent Arena SteerabilityAgents9.64.1−2.7
ACCoRDGeneral QA77.474.7−2.7
CursorBench 3.2Code62.364.9+2.6
DeepResearch Bench II (Mercor 130-task set)Search44.647.2+2.6
EnigmaEvalGeneral QA20.523.1+2.6
WMDP-Bio (SecureBio)Science92.389.7−2.6
MortgageTax (Vals all 1,258 documents)Finance & Accounting69.967.3−2.6
LMCAGeneral QA57.555.0−2.6
HardGeoBench <=10miMultimodal27.525.0−2.5
GDPval-AA v2Professional Work15291479−2.5
LMArena Text Arena EloChat & Writing14821468−2.4
Legal Research BenchLegal43.841.3−2.4
Artificial Analysis Coding IndexIndexes74.376.7+2.4
LAB-Bench LitQA2 (SecureBio)Science48.751.1+2.4
Design Arena Overall Code CategoriesCode12521238−2.3
MindCube TinyMultimodal64.962.6−2.3
GDP.pdfProfessional Work22.524.7+2.2
Finance Agent v2 - ComparablesFinance & Accounting44.947.0+2.2
ChessBench EloGames961925−2.1
AutomationBench 1.0.6 (Zapier)Agents16.914.8−2.1
LiveCodeBench (Vals Public) - MediumCode89.887.7−2.1
GPQA DiamondGeneral QA91.989.9−2.0
Vals Index v2.1Indexes55.153.1−2.0
AutomationBench (Overall)Agents17.215.2−2.0
DiligenceBenchFinance & Accounting47.149.1+2.0
ProofBenchMath69.071.0+2.0
ZeroBench (Main questions, pass@5)Multimodal17.019.0+2.0
Artificial Analysis Progress Index v1Indexes318324+1.9
MMLU ProGeneral QA90.088.1−1.9
LiveCodeBench (Vals Public) - EasyCode99.197.2−1.9
ALE-BenchCode21022154+1.9
LiveBench 2.0Agents76.875.0−1.8
AIME 2025Math98.3100.0+1.7
MRCR v2 (8-needle, Context Arena) - 128K AUCGeneral QA87.889.4+1.6
DTBenchGeneral QA94.993.3−1.6
LegalBenchLegal83.685.1+1.5
FACTS Benchmark SuiteGeneral QA53.552.0−1.5
ObviousBenchGeneral QA99.397.9−1.4
OTIS Mock AIME 2024-2025Math98.399.7+1.4
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting66.467.8+1.4
Ruby LLM Benchmarks — Program FixerCode72.674.0+1.3
Finance Agent v2 - AdjustmentsFinance & Accounting48.449.7+1.3
Finance Agent v2 (Vals Index subset)Finance & Accounting53.952.6−1.3
Finance Agent v2 (Vals Multimodal Index component)Finance & Accounting53.952.6−1.3
Vals Index v2Indexes60.959.6−1.3
Ramp SWE-Bench (78 published tasks)Code78.276.9−1.3
Finance Agent v2 - Disclosure AnalysisFinance & Accounting65.766.9+1.2
CorpFin v2 - Exact PagesFinance & Accounting64.363.2−1.2
MASKSafety94.493.3−1.1
SciCode (main problems)Science16.915.8−1.1
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting71.672.7+1.0
ARC-AGI-3 (Semi-Private)General QA1.50.5−1.0
Mystery Game PuzzlesGames36.035.0−1.0
VCT (SecureBio)Science48.247.2−1.0
Wiz Cyber Model Arena v3.9Cybersecurity65.464.4−1.0
EpiBenchScience39.039.9+0.9
CharXiv (descriptive and reasoning combined)Multimodal93.294.1+0.9
AA-Briefcase EloAgents13161330+0.8
SWE Atlas - Codebase QnACode44.445.2+0.8
Humanity's Sixth SenseMultimodal23.522.7−0.8
BioSecBench-SurveillanceScience50.251.0+0.8
Finance Agent v2 - PrecedentsFinance & Accounting25.624.8−0.8
Epoch Capabilities Index (ECI)Indexes160161+0.8
Finance Agent v2 - All PassFinance & Accounting43.542.8−0.8
Terminal-Bench HardCode58.357.6−0.8
Diligence Stack Agent BenchFinance & Accounting77.376.7−0.6
MedXpertQA MM Extended (Mercor held-out)Health & Medicine51.651.0−0.6
SciCode 1.0.1 (Artificial Analysis)Code54.455.0+0.6
Code MigrationCode47.347.8+0.6
ChessBench AccuracyGames72.471.9−0.5
TaxEval v2Finance & Accounting75.676.2+0.5
Terminal-Bench 4.0Code23.222.7−0.5
MBCT (SecureBio)Science56.456.9+0.5
ProgramBenchCode1.00.5−0.5
CorpFin v2 - Shared Max ContextFinance & Accounting66.066.4+0.5
CorpFin v2 (Vals Index component)Finance & Accounting66.066.4+0.5
CorpFin v2 (Vals Multimodal Index component)Finance & Accounting66.066.4+0.5
TextQuests Game ProgressGames40.140.5+0.4
GDPval-AA v2 (normalized)Professional Work54.554.1−0.4
SkillsBench with SkillsAgents59.258.9−0.3
Finance Agent v2 - Financial ModelingFinance & Accounting23.323.0−0.3
Artificial Analysis Intelligence Index v4.3.2Indexes41.842.1+0.3
Tax Agent BenchFinance & Accounting64.965.2+0.3
Artificial Analysis Intelligence Index v4.3Indexes42.042.3+0.3
Artificial Analysis Agentic IndexIndexes47.247.4+0.2
MMLU-Bio (SecureBio)Science90.990.7−0.2
SEA-HELMGeneral QA75.675.4−0.2
LAB-Bench ProtocolQA (SecureBio)Science76.977.0+0.1
SpatialViz-BenchMultimodal65.265.20.0
GDPval-AA v2.1Professional Work146114610.0
CadQueryEvalCode56.056.00.0
CadQueryEval (Corrected Geometry)Code68.068.00.0
HardGeoBench <=100miMultimodal36.336.30.0
Harvey's Legal Agent Benchmark - Antitrust CompetitionLegal0.00.00.0
Harvey's Legal Agent Benchmark - Arbitration International Dispute ResolutionLegal0.00.00.0
Harvey's Legal Agent Benchmark - Banking FinanceLegal0.00.00.0
Harvey's Legal Agent Benchmark - Bankruptcy RestructuringLegal0.00.00.0
Harvey's Legal Agent Benchmark - Corporate GovernanceLegal0.00.00.0
Harvey's Legal Agent Benchmark - Corporate M&ALegal0.00.00.0
Harvey's Legal Agent Benchmark - Employment LaborLegal0.00.00.0
Harvey's Legal Agent Benchmark - Environmental ESGLegal0.00.00.0
Harvey's Legal Agent Benchmark - Funds Asset ManagementLegal0.00.00.0
Harvey's Legal Agent Benchmark - ImmigrationLegal0.00.00.0
Harvey's Legal Agent Benchmark - InsuranceLegal0.00.00.0
Harvey's Legal Agent Benchmark - Intellectual PropertyLegal0.00.00.0
Harvey's Legal Agent Benchmark - International Trade SanctionsLegal0.00.00.0
Harvey's Legal Agent Benchmark - Litigation Dispute ResolutionLegal0.00.00.0
Harvey's Legal Agent Benchmark - Real EstateLegal0.00.00.0
Harvey's Legal Agent Benchmark - Structured Finance SecuritizationLegal20.020.00.0
Harvey's Legal Agent Benchmark - TaxLegal0.00.00.0
Harvey's Legal Agent Benchmark - Trusts Estates Private ClientLegal0.00.00.0
Harvey's Legal Agent Benchmark - White Collar Defense InvestigationsLegal0.00.00.0
Terminal-Bench Science 0.1 (Vals)Science10.010.00.0

Opus 4.8vsGPT-5.6 Terranormalised score marginsData Nov 15, 2026Data: Vals AI, Mercor, Artificial Analysis and 69 othersVillage LLM Atlasatlas.village.dev

Win counts use each benchmark’s declared score direction. Represented suite slices and the derived Capability score are excluded from headline totals. Missing or incompatible setups are excluded, never counted as losses.
Loading Atlas data…