Atlas

Compare Models

DeepSeek V4.1 Flash 1017 vs MiniMax M3.1 Flash

DeepSeek V4.1 Flash 1017 leads by 1.1 capability points (≈95% interval −1.5 to +3.7): statistically tied · P(higher) ≈ 80%

Wins 34 of 39 shared benchmarks (5 lost)

Capability vs. relative cost
DeepSeek V4.1 Flash 1017MiniMax 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.

DeepSeek V4.1 Flash 1017vsMiniMax M3.1 Flashcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins34 – 5 · 39 shared

Comparing matching setups only (58 benchmarks excluded) · suites counted once
← DeepSeek V4.1 Flash 1017MiniMax M3.1 Flash →

DeepSeek V4.1 Flash 1017vsMiniMax M3.1 Flashwins by categoryData Nov 15, 2026Data: Vals AI, Mercor, DeepSeek and 19 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 DeepSeek V4.1 Flash 1017.

FORTRESS Average Risk Score (ARS)Safety27.252.1−24.9
NoCoT-Bench NCKI (kspine v3)General QA10265.9−24.7
Visa Agentic Commerce BenchmarkAgents57.833.3−24.4
Code Migration v1.1 (Vals Index v2.2 subset)Code68.548.2−20.3
ALE-BenchCode20121495−18.3
VoxelBench Text RatingGames20471810−16.9
DAXBench v2Code77.763.8−13.8
Vibe Code Bench v1.1Code87.773.9−13.8
Hebbia Financial Services BenchmarkFinance & Accounting60.347.0−13.3
FoodTruck BenchGames5328963464+13.1
GBENCH One-shot Coding GScoreGames53.841.1−12.6
Tax Agent BenchFinance & Accounting74.962.7−12.2
APEX-Agents 1.1Professional Work79.468.1−11.3
Legal Research BenchLegal45.234.1−11.1
o11y-bench 1.1 Pass^3Agents75.064.1−10.9
Vals Index v2.2Indexes61.353.8−7.5
NoCoT-Bench NCRI 15.2General QA81.668.4−7.2
NL2Repo - release-reported repository generation scoreCode68.261.2−7.0
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal18.811.9−6.9
HLAB-Vals-1.1Legal14.27.5−6.7
Terminal-Bench 4.0Code31.825.3−6.6
Agents' Last Exam v1.1 (ALE v1.1)Agents36.129.5−6.6
LiveBench 2.0Agents79.273.1−6.1
τ³-Banking (Mercor run, bm25_grep retrieval)Agents41.636.1−5.5
Vals Finance IndexIndexes60.955.9−5.0
Vals Legal IndexIndexes71.967.1−4.8
Finance Agent v2.1Finance & Accounting62.958.8−4.1
Humanity's Last Exam (Text-Only)General QA44.048.0+4.0
APEX-1 (Mercor, mean score)Professional Work68.365.0−3.3
DeepSWE v1.1Code77.073.7−3.3
BabyVision with toolsMultimodal92.591.0−1.5
GDPxProfessional Work5.44.0−1.4
Excel Modeling Benchmark v1.1Finance & Accounting69.267.8−1.3
Design Arena Overall Code CategoriesCode12891297+1.3
BuseyBench SVGMultimodal4.34.2−1.3
ErdosBenchMath3.03.0−0.7
APEX-Accounting 1.1Finance & Accounting59.360.0+0.7
FrontierML ($200 wallet)Agents23.523.2−0.2
Access to Justice BenchmarkLegal73.173.3+0.2

DeepSeek V4.1 Flash 1017vsMiniMax M3.1 Flashnormalised score marginsData Nov 15, 2026Data: Vals AI, Mercor, DeepSeek and 19 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…