Atlas

Compare Models

Gemini 3.7 Flash vs GPT-5.5

GPT-5.5 leads by 0.6 capability points (≈95% interval −1.3 to +2.6): statistically tied · P(higher) ≈ 74%

Gemini 3.7 Flash wins 106 of 221 shared benchmarks (93 lost, 22 tied)

Capability vs. relative cost
Gemini 3.7 FlashGPT-5.5Closed 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.

Gemini 3.7 FlashvsGPT-5.5capability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins106 – 93 · 221 shared, 22 tied

Comparing matching setups only (57 benchmarks excluded) · suites counted once
← Gemini 3.7 FlashGPT-5.5 →

Gemini 3.7 FlashvsGPT-5.5wins by categoryData Nov 15, 2026Data: Vals AI, Mercor, Artificial Analysis and 45 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-5.5 minus Gemini 3.7 Flash.

BrokenArXiv 06/2026Math11.169.4+58.3
Olam Social Poker EloGames14681521+57.6
ATLAS-Education (StudentBench) Lesson PlanningProfessional Work-0.40.9+53.4
BrokenArXiv 04/2026Math19.372.1+52.9
BrokenArXiv 05/2026Math8.550.0+41.5
ChessBench — Historical Output RepairGames70.729.3−41.3
MASKSafety53.790.1+36.4
FrontierMath Tier 4 (v2) - PrivateMath36.672.5+35.9
LLM Creative Story WritingChat & Writing-0.72.4+35.5
HieroglyphBenchMultimodal55.022.8−32.2
MirrorCode (ML, +Private, 2L)Code41.110.0−31.1
ArXivMath 05/2026Math47.577.5+30.0
HardGeoBench Country AccuracyMultimodal67.545.0−22.5
Antidote: Everyday EditionSafety10751013−22.1
AA-LCR Extended (Mercor held-out)Search48.970.5+21.6
ArXivMath 06/2026Math62.583.6+21.1
Harvey's Legal Agent Benchmark - Banking FinanceLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Capital MarketsLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Corporate GovernanceLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - Data Privacy CybersecurityLegal0.020.0+20.0
Harvey's Legal Agent Benchmark - Energy Natural ResourcesLegal40.020.0−20.0
Harvey's Legal Agent Benchmark - Healthcare Life SciencesLegal20.00.0−20.0
Harvey's Legal Agent Benchmark - InsuranceLegal20.00.0−20.0
Hemingway-benchChat & Writing11051035−19.4
Mystery Game PuzzlesGames37.056.0+19.0
LMArena Agent Arena Net ImprovementAgents-1.34.2+18.3
HardGeoBench <=10miMultimodal46.328.8−17.5
Furniture Assembly BenchmarkMultimodal26.744.2+17.5
LLM Chess EloGames11071532+17.4
ChessBench CoherenceGames83.2100.0+16.8
TxBench Antibody Discovery V0Science50.033.6−16.4
HardGeoBench <=100miMultimodal56.340.0−16.2
RiemannBenchMath39.255.2+16.0
HPCT Refusal (CAIS)Safety13.029.0+16.0
SM BenchSafety80.665.4−15.2
Senior SWE-BenchCode14.729.5+14.8
AutomationBench-AA 1.0.6Agents62.047.3−14.7
Finance Agent v2 - Market AnalysisFinance & Accounting74.560.3−14.2
NoCoT-Bench NCRI 15.2General QA126100−14.2
Roboflow Vision Evals (October 2026, five tasks)Multimodal80.166.0−14.1
FrontierMath Tiers 1-3 (v2) - PrivateMath71.685.3+13.7
TxBench Antibody Discovery V1 Clean RerunsScience43.529.9−13.6
Finance Agent v2 - PrecedentsFinance & Accounting33.821.9−11.9
Machiavelli Deception (CAIS)Safety85.581.7+11.5
DeepSWE v1.1Code65.554.0−11.5
ArXivMath 04/2026Math57.568.8+11.3
VCT-Refusal (CAIS)Safety4.816.0+11.2
CritPt (avg@5)Science14.325.4+11.1
BioTIER RefusalSafety84.473.4−11.0
ZeroBench (Main questions, pass@5, with tools)Multimodal30.041.0+11.0
HANDBOOK.mdProfessional Work10.821.5+10.7
SpatialBench-Long V0 VerifiedScience22.733.3+10.6
Roboflow Vision Evals (September 2026 protocol)Multimodal85.274.8−10.4
Code MigrationCode34.845.2+10.4
LLM Debate BenchmarkChat & Writing14551528+10.3
TextQuests Average Cumulative HarmSafety24.421.4+10.2
AA-AnalystAgent (pass^5)Professional Work60.050.0−10.0
TaxCalcBench TY25Finance & Accounting14.024.0+10.0
GBENCH One-shot Coding GScoreGames52.662.2+9.6
ChartographyMultimodal40.431.0−9.4
MRCR v2 (8-needle, Context Arena) - 128K pointwiseGeneral QA88.679.3−9.3
SWE-bench Verified Extended (Mercor held-out)Code30.039.3+9.3
HardGeoBench Avg DistanceMultimodal8951196−9.3
SpeciEval (Pooled English Runs)Safety89.098.2+9.2
SpeciEval (Refusal-Aware Retries, Historical)Safety89.098.2+9.2
APEX-Agents 1.1Professional Work79.470.3−9.1
Finance Agent v2 - AdjustmentsFinance & Accounting54.745.7−9.0
LAB-Bench ProtocolQA (SecureBio)Science86.877.9−8.9
Finance Agent v2 - Disclosure AnalysisFinance & Accounting67.758.9−8.8
Harvey HLAB — Criteria Pass RateLegal88.680.0−8.6
MindCube TinyMultimodal88.479.9−8.5
IntPhys 2Multimodal68.259.9−8.3
ALE-BenchCode18762108+8.2
Finance Agent v2 - General Quantitative AnalysisFinance & Accounting81.873.7−8.1
CadQueryEval (Corrected Geometry)Code84.076.0−8.0
Finance Agent v2 - All PassFinance & Accounting47.539.6−7.9
Design Arena Overall Code CategoriesCode13041257−7.8
LMArena Agent Arena Bash RecoveryAgents-5.49.5+7.5
MMMU-Pro Extended (Mercor held-out)Multimodal36.443.8+7.4
ERQAMultimodal77.770.5−7.2
NoCoT-Bench NCKI (kspine v3)General QA121111−7.2
ComplexConstraintsChat & Writing42.449.5+7.1
Document ArenaProfessional Work14761484+7.0
MedXpertQA MM (200-question set)Health & Medicine83.576.7−6.8
Excel Modeling BenchmarkFinance & Accounting71.364.5−6.8
ScBench V1 Clean RerunsScience50.457.1+6.7
Finance Agent v2 - ComparablesFinance & Accounting50.343.7−6.6
BrowseComp Extended (Mercor held-out)Search25.519.0−6.5
Terminal-Bench 2.1 Extended (Mercor held-out)Code44.438.0−6.4
BuseyBench SVGMultimodal6.66.0−6.4
Terminal-Bench 2.1Code85.879.4−6.4
Creative Writing v3 EloChat & Writing17201845+6.3
EnterpriseOps-Gym-AAAgents50.444.1−6.3
EpiBench V1 Clean RerunsScience22.028.3+6.3
SimpleQA Verified (Epoch anti-abstention)General QA69.263.0−6.2
GeneBench-Pro Public Release SubsetScience14.08.0−6.0
SUPERChemScience72.866.8−6.0
TxBench Oligo Discovery V0Science39.533.6−5.9
Finance Agent v2 - Earnings AnalysisFinance & Accounting79.173.2−5.9
GPQA-Bio (SecureBio)Science81.687.4+5.8
Legal Research BenchLegal34.640.4+5.8
MRCR v2 (8-needle, Context Arena) - 128K AUCGeneral QA93.988.2−5.7
BioSecBench-Surveillance V1 Clean RerunsScience41.547.1+5.6
AHK-EvalCode91.797.2+5.6
APEX-SWECode42.036.6−5.4
BioSecBench-Function V1 Clean RerunsScience33.728.3−5.4
Harvey LAB (Mercor run, 1,749 public tasks)Legal9.34.1−5.2
PCT Political Inconsistency (CAIS)Safety45.140.0+5.1
ChessBench EloGames12651349+5.0
SkillsBench without SkillsAgents44.949.9+5.0
SpatialBench V1 Clean RerunsScience59.164.1+4.9
MineBench — Bradley-Terry RatingGames17651846+4.9
SWE-bench MultilingualCode82.677.7−4.9
EnterpriseBench: CoreCraftProfessional Work55.951.3−4.6
Terminal-Bench 4.0 (Artificial Analysis)Code13.69.1−4.5
Judgemark v4Chat & Writing83.487.8+4.4
LAB-Bench LitQA2 (SecureBio)Science54.249.8−4.4
MedCodeHealth & Medicine53.449.1−4.3
TextQuests Game ProgressGames52.948.9−4.0
LMCAGeneral QA50.454.3+3.9
Conceptual Reasoning IndexIndexes64.667.3+3.8
SciCode 1.0.1 (Artificial Analysis)Code59.856.1−3.7
MMMU ProMultimodal84.981.2−3.7
SkillsBench with SkillsAgents65.962.2−3.7
TxBench Oligo Discovery V1 Clean RerunsScience35.631.9−3.6
CharXiv Extended (Mercor held-out)Multimodal79.182.7+3.6
BioSecBench-Function V0Science40.136.5−3.6
ACCoRDGeneral QA77.080.5+3.5
𝜏³-BankingAgents35.539.0+3.5
LMArena Agent Arena SteerabilityAgents-0.06.9+3.4
GlobeBenchGeneral QA21.825.2+3.4
LiveCodeBench (Vals AI public dataset)Code88.785.3−3.4
Creative Writing v3 Rubric ScoreChat & Writing15.215.9+3.4
EnigmaEvalGeneral QA33.937.2+3.3
VCT v2 (SecureBio)Science48.645.3−3.3
AIME 2025Math96.7100.0+3.3
GDPval Extended (Mercor held-out rubric score)Professional Work57.854.5−3.3
Humanity's Last Exam DiamondGeneral QA31.227.9−3.3
Bio Capabilities Index (SecureBio)Indexes162158−3.2
Agent Red Teaming k=100 (CAIS)Safety38.541.5−3.1
GDPval-AA v2.1Professional Work14041335−3.0
GDPval-AA v2Professional Work14331372−3.0
VCT (SecureBio)Science49.352.3+3.0
Chess Puzzles (task version 2.0)Games66.063.0−3.0
AA-Omniscience IndexIndexes26.520.5−3.0
MedScribeHealth & Medicine83.986.9+2.9
ScBench-Long V1 Clean RerunsScience18.821.6+2.8
Tax Agent BenchFinance & Accounting57.760.5+2.8
Medical Long Context Reasoning (MLCR-AA)Health & Medicine15.017.8+2.8
OTIS Mock AIME 2024-2025Math97.2100.0+2.8
Vision Arena OverallMultimodal12941283−2.7
SpatialBench-Long V1 Clean RerunsScience29.832.1+2.4
SAGEMath49.251.5+2.3
Toloka Arena (v1)Agents54.652.3−2.3
GDP.pdfProfessional Work23.826.0+2.2
MortgageTax (Vals all 1,258 documents)Finance & Accounting66.768.8+2.1
LAB-Bench CloningScenarios (SecureBio)Science99.497.3−2.1
MBCT (SecureBio)Science55.057.1+2.1
HPCT (SecureBio)Science64.766.8+2.1
DeepResearch Bench II (Mercor 130-task set)Search43.445.4+2.0
MedXpertQA MM Extended (Mercor held-out)Health & Medicine62.760.8−1.9
Vals Index v2Indexes59.357.4−1.9
Humanity's Last ExamGeneral QA45.643.8−1.8
SWE-bench VerifiedCode80.882.6+1.8
WCB (SecureBio)Science58.059.7+1.7
MMLU ProGeneral QA90.688.9−1.7
ForecastBench Preliminary Dataset Brier IndexForecasting61.763.3+1.6
AA-Briefcase EloAgents11161089−1.6
LiveBench 2.0Agents67.268.7+1.5
ScBench-Long V0 VerifiedScience24.222.7−1.5
BrowseComp (Mercor 130-question subset)Search79.277.7−1.5
LMArena Text Arena EloChat & Writing14891481−1.4
APEX-1 (Mercor, mean score)Professional Work67.068.4+1.4
AA-LCR v1.1Search83.084.3+1.3
Epoch Capabilities Index (ECI)Indexes158160+1.3
Finance Agent v2 - Financial ModelingFinance & Accounting24.222.9−1.2
RedlineBenchLegal45.344.1−1.2
SpatialViz-BenchMultimodal75.374.2−1.1
Artificial Analysis Progress Index v1Indexes317320+0.9
APEX-Accounting 1.1Finance & Accounting57.256.3−0.9
SpatialBench V0 VerifiedScience64.365.2+0.9
VariantBench V1 Clean RerunsScience31.430.5−0.8
Artificial Analysis Intelligence Index v4.3Indexes39.438.6−0.8
DTBenchGeneral QA96.896.0−0.8
ForecastBench Tournament Overall Brier IndexForecasting66.267.0+0.8
LegalBenchLegal87.386.5−0.7
Artificial Analysis Intelligence Index v4.3.2Indexes39.138.4−0.7
CharXiv (descriptive and reasoning combined)Multimodal95.394.6−0.7
GDPval (Mercor 213-task rubric score)Professional Work85.284.6−0.6
Vibe Code Bench v1.1Code70.469.8−0.5
ARC-AGI-1 Semi-PrivateGeneral QA95.595.0−0.5
ProgramBenchCode0.00.5+0.5
ChessBench AccuracyGames79.278.8−0.4
ARC-AGI-2General QA84.685.0+0.4
ATLAS-Education (StudentBench) Learning GainProfessional Work14.113.5−0.3
Finance Agent v2 - General Qualitative AnalysisFinance & Accounting65.365.6+0.3
TaxEval v2Finance & Accounting74.775.0+0.2
GPQA DiamondGeneral QA95.595.50.0
PG-LLM-50Science0.30.30.0
BabaIsBench Stage 0: The IntroGames100.0100.00.0
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 - Bankruptcy RestructuringLegal0.00.00.0
Harvey's Legal Agent Benchmark - Corporate M&ALegal0.00.00.0
Harvey's Legal Agent Benchmark - Emerging Companies Venture CapitalLegal0.00.00.0
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 - 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 SecuritizationLegal0.00.00.0
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
ObviousBenchGeneral QA100.0100.00.0
OncologyBench Strict (25 tasks, October 9 rubric)Health & Medicine14.714.70.0
TxBench Preclinical Pharmacology V1 Clean RerunsScience55.355.30.0
WordleBenchGames97.097.00.0

Gemini 3.7 FlashvsGPT-5.5normalised score marginsData Nov 15, 2026Data: Vals AI, Mercor, Artificial Analysis and 45 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…