Atlas

Compare Models

Gemini 4 Carbon vs GPT-6.1 Sol

Gemini 4 Carbon leads by 1.7 capability points (≈95% interval −1.1 to +4.4): statistically tied · P(higher) ≈ 88%

Wins 50 of 60 shared benchmarks (7 lost, 3 tied)

Capability vs. relative cost
Gemini 4 CarbonGPT-6.1 SolClosed 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 4 CarbonvsGPT-6.1 Solcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins50 – 7 · 60 shared, 3 tied

Comparing matching setups only (23 benchmarks excluded) · suites counted once
← Gemini 4 CarbonGPT-6.1 Sol →

Gemini 4 CarbonvsGPT-6.1 Solwins by categoryData Nov 15, 2026Data: Artificial Analysis, Vals AI, Arena and 8 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-6.1 Sol minus Gemini 4 Carbon.

TxBench-Antibody-Development V1 (60 evaluations)Science61.127.8−33.3
Terminal-Bench-Science 0.1Science62.437.1−25.2
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine19.438.3+18.9
Legal Research BenchLegal55.838.5−17.3
APEX-Agents 1.1Professional Work90.473.2−17.2
TxBench Antibody Discovery V1 (100 evaluations)Science65.049.7−15.3
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal23.08.8−14.2
Tax Agent BenchFinance & Accounting76.462.3−14.1
τ³-Banking (Mercor run, bm25_grep retrieval)Agents55.743.0−12.7
Terminal-Bench 4.0 (Artificial Analysis)Code60.648.0−12.6
HLAB-Vals-1.1Legal17.56.3−11.2
AutomationBench-AA 1.0.6Agents77.866.6−11.2
Vals Finance IndexIndexes67.055.8−11.2
APEX-1 (Mercor, mean score)Professional Work80.069.3−10.7
Humanity's Last Exam DiamondGeneral QA69.158.8−10.3
GDPval-AA v2.1Professional Work17331521−9.3
LMArena Agent Arena Net Improvement (Steering Burden)Agents17.012.5−8.5
OpenWorldEngineer (Mean@8)Code32.841.2+8.4
Artificial Analysis Legal IndexIndexes62.053.7−8.3
AA-Briefcase v1.1 EloAgents16111469−7.8
LMArena Text Arena EloChat & Writing15261482−7.7
Vals Legal IndexIndexes75.968.7−7.2
LMArena Agent Arena Praise vs ComplaintAgents41.027.5−6.8
MKA (Artificial Analysis)Professional Work54.347.7−6.6
APEX-Accounting 1.1Finance & Accounting70.764.3−6.4
Artificial Analysis Finance & Accounting IndexIndexes62.656.2−6.4
TxBench Preclinical Pharmacology V1 Clean RerunsScience66.059.7−6.3
Artificial Analysis Engineering IndexIndexes65.359.4−6.0
AA-Omniscience v1.1 IndexIndexes53.542.4−5.6
Artificial Analysis Intelligence Index v5.0.1Indexes60.054.6−5.4
Excel Modeling Benchmark v1.1Finance & Accounting76.571.1−5.4
Finance Agent v2.1Finance & Accounting67.261.8−5.3
LMArena Agent Arena Bash ReliabilityAgents-5.25.0+5.1
VIALS-AA (pass^3)Multimodal64.059.0−5.0
Terminal-Bench 4.0Code63.158.2−5.0
Code Migration v1.1 (Vals Index v2.2 subset)Code80.175.1−4.9
Artificial Analysis Healthcare & Medical IndexIndexes56.852.0−4.8
AA-LCR v1.1Search79.384.0+4.7
MMMU ProMultimodal89.285.1−4.2
FrontierCode 2.0 (Artificial Analysis)Code40.636.7−3.8
LMArena Agent Arena Confirmed SuccessAgents18.411.7−3.3
Artificial Analysis Strategy & Ops IndexIndexes60.957.6−3.3
Artificial Analysis Economics IndexIndexes59.656.3−3.3
TxBench Oligo Discovery V1 (113 evaluations)Science55.852.5−3.2
Artificial Analysis Intelligence Index v5.0Indexes57.754.5−3.2
LMArena Agent Arena Steering BurdenAgents19.113.4−2.8
VIALS-AA (average of 3)Multimodal67.364.8−2.5
GDPxProfessional Work10.98.4−2.4
GDP.pdf (Artificial Analysis, All-pass)Professional Work29.632.0+2.4
ARC-AGI-3 (Semi-Private)General QA96.995.0−1.9
DAXBench v2Code93.692.6−1.1
BuseyBench SVGMultimodal7.98.0+0.9
Artificial Analysis Progress Index v1Indexes351348−0.8
RiemannBench (November 2026 rescoring)Math80.079.2−0.8
CritPt v2 (agentic, avg@5)Science82.382.0−0.3
LMArena Agent Arena Tool HallucinationAgents0.51.0+0.2
Vibe Code Bench v1.1Code89.188.9−0.1
ARC-AGI-1 Semi-PrivateGeneral QA98.598.50.0
ARC-AGI-2General QA94.294.20.0
VIALS-AA (pass@3)Multimodal70.870.80.0

Gemini 4 CarbonvsGPT-6.1 Solnormalised score marginsData Nov 15, 2026Data: Artificial Analysis, Vals AI, Arena and 8 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…