Atlas

Compare Models

Claude Opus 5 vs MiniMax M3.1 Flash

Claude Opus 5 leads by 4.2 capability points (≈95% interval +1.7 to +6.7): a clear lead over MiniMax M3.1 Flash · P(higher) ≈ 100%

Wins 24 of 27 shared benchmarks (3 lost)

Capability vs. relative cost
Claude Opus 5MiniMax 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.

Claude Opus 5vsMiniMax M3.1 Flashcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins24 – 3 · 27 shared

Comparing matching setups only (46 benchmarks excluded) · suites counted once
← Claude Opus 5MiniMax M3.1 Flash →

Claude Opus 5vsMiniMax M3.1 Flashwins by categoryData Nov 15, 2026Data: Mercor, MiniMax, Vals AI 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 Claude Opus 5.

Toloka Arena (v3)Agents62.022.0−40.0
ALE-BenchCode26221495−39.9
NoCoT-Bench NCKI (kspine v3)General QA11965.9−37.0
GBENCH One-shot Coding GScoreGames68.141.1−27.0
NoCoT-Bench NCRI 15.2General QA11568.4−25.7
DAXBench v2Code86.263.8−22.3
Legal Research BenchLegal55.334.1−21.2
BuseyBench SVGMultimodal6.34.2−20.8
TaxCalcBench TY25Finance & Accounting20.00.0−20.0
Vibe Code Bench v1.1Code88.473.9−14.5
NL2Repo - release-reported repository generation scoreCode75.361.2−14.1
Tax Agent BenchFinance & Accounting76.762.7−13.9
AutomationBench (Public Set)Agents50.362.8+12.5
GDPval-AA v2.1Professional Work17081465−10.7
Humanity's Last Exam DiamondGeneral QA38.628.1−10.5
LiveBench 2.0Agents82.173.1−9.0
APEX-Agents 1.1Professional Work77.068.1−8.9
HiL-Bench (Human-in-Loop Benchmark)Agents57.065.7+8.7
τ³-Banking (Mercor run, bm25_grep retrieval)Agents44.036.1−7.9
PostTrainBench Average (v1.1)Indexes45.639.3−6.3
APEX-Accounting 1.1Finance & Accounting64.360.0−4.3
Design Arena Overall Code CategoriesCode13211297−4.0
APEX-1 (Mercor, mean score)Professional Work68.665.0−3.6
BabyVision with toolsMultimodal94.191.0−3.1
Agents' Last Exam (ALE)Agents32.229.6−2.6
CharXiv ReasoningMultimodal83.782.5−1.2
DeepSWE v1.1Code73.273.7+0.5

Claude Opus 5vsMiniMax M3.1 Flashnormalised score marginsData Nov 15, 2026Data: Mercor, MiniMax, Vals AI 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…