Atlas

Compare Models

Gemini 4 Argon vs MiniMax M3.1 Flash

Gemini 4 Argon leads by 6.1 capability points (≈95% interval +3.5 to +8.6): a clear lead over MiniMax M3.1 Flash · P(higher) ≈ 100%

Wins 35 of 37 shared benchmarks (2 lost)

Capability vs. relative cost
Gemini 4 ArgonMiniMax M3.1 FlashClosed 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 ArgonvsMiniMax M3.1 Flashcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins35 – 2 · 37 shared

Comparing matching setups only (57 benchmarks excluded) · suites counted once
← Gemini 4 ArgonMiniMax M3.1 Flash →

Gemini 4 ArgonvsMiniMax M3.1 Flashwins by categoryData Nov 15, 2026Data: Vals AI, Scale AI, Mercor and 17 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is MiniMax M3.1 Flash minus Gemini 4 Argon.

Toloka Arena (v3)Agents73.022.0−51.0
Visa Agentic Commerce BenchmarkAgents82.233.3−48.9
ALE-BenchCode28051495−46.4
NoCoT-Bench NCRI 15.2General QA15168.4−45.5
NoCoT-Bench NCKI (kspine v3)General QA12865.9−43.3
BuseyBench SVGMultimodal7.64.2−33.4
Terminal-Bench 4.0Code57.625.3−32.3
GBENCH One-shot Coding GScoreGames67.941.1−26.7
Code Migration v1.1 (Vals Index v2.2 subset)Code74.448.2−26.2
Humanity's Last Exam DiamondGeneral QA53.228.1−25.1
DAXBench v2Code88.363.8−24.5
τ³-Banking (Mercor run, bm25_grep retrieval)Agents58.136.1−22.0
Legal Research BenchLegal54.834.1−20.7
HLAB-Vals-1.1Legal27.17.5−19.6
APEX-Agents 1.1Professional Work87.468.1−19.3
o11y-bench 1.1 Pass^3Agents81.364.1−17.2
Tax Agent BenchFinance & Accounting79.062.7−16.3
FrontierML ($200 wallet)Agents39.523.2−16.3
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal27.411.9−15.5
APEX-Accounting 1.1Finance & Accounting71.560.0−11.5
LiveBench 2.0Agents84.173.1−11.0
Agents' Last Exam v1.1 (ALE v1.1)Agents40.229.5−10.7
Vals Finance IndexIndexes66.055.9−10.1
Agents' Last Exam (ALE)Agents39.529.6−9.9
DeepSWE v1.1Code77.968.1−9.8
Vals Legal IndexIndexes76.867.1−9.7
Finance Agent v2.1Finance & Accounting67.758.8−8.8
Excel Modeling Benchmark v1.1Finance & Accounting74.867.8−7.0
Access to Justice BenchmarkLegal68.573.3+4.8
Design Arena Overall Code CategoriesCode13261297−4.8
Professional Reasoning Benchmark - LegalLegal49.445.1−4.3
Professional Reasoning Benchmark - FinanceFinance & Accounting49.945.6−4.2
SAGE v1.1Math89.585.3−4.2
HiL-Bench (Human-in-Loop Benchmark)Agents62.065.7+3.7
GDPxProfessional Work7.34.0−3.3
SWE-Bench Pro V2 — FullCode99.296.6−2.7
ErdosBenchMath3.03.0−0.7

Gemini 4 ArgonvsMiniMax M3.1 Flashnormalised score marginsData Nov 15, 2026Data: Vals AI, Scale AI, Mercor and 17 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…