Atlas

Compare Models

Claude Haiku 5.5 vs MiniMax M3.1 Flash

Claude Haiku 5.5 leads by 1.5 capability points (≈95% interval −0.9 to +3.8): statistically tied · P(higher) ≈ 89%

Wins 24 of 33 shared benchmarks (9 lost)

Capability vs. relative cost
Claude Haiku 5.5MiniMax 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 Haiku 5.5vsMiniMax M3.1 Flashcapability vs. relative costData Nov 15, 2026Village LLM Atlasatlas.village.dev

Benchmark wins24 – 9 · 33 shared

Comparing matching setups only (57 benchmarks excluded) · suites counted once
← Claude Haiku 5.5MiniMax M3.1 Flash →

Claude Haiku 5.5vsMiniMax M3.1 Flashwins by categoryData Nov 15, 2026Data: Vals AI, Mercor, Artificial Analysis and 13 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 Haiku 5.5.

o11y-bench 1.1 Pass^3Agents85.964.1−21.9
DAXBench v2Code81.963.8−18.1
ALE-BenchCode19961495−17.7
Vibe Code Bench v1.1Code90.473.9−16.5
GBENCH One-shot Coding GScoreGames55.641.1−14.5
NoCoT-Bench NCKI (kspine v3)General QA85.965.9−13.9
FrontierML ($200 wallet)Agents34.923.2−11.7
Terminal-Bench 4.0Code36.425.3−11.1
Legal Research BenchLegal43.834.1−9.6
SWE-Bench Pro V2 — FullCode87.296.6+9.3
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Legal4.911.9+7.0
τ³-Banking (Mercor run, bm25_grep retrieval)Agents43.036.1−6.9
GDPval-AA v2.1Professional Work16201465−6.8
Finance Agent v2.1Finance & Accounting65.458.8−6.5
AA-Briefcase v1.1 EloAgents15781474−5.7
Vals Index v2.2Indexes58.753.8−4.9
SAGE v1.1Math89.885.3−4.6
BuseyBench SVGMultimodal4.64.2−3.4
HLAB-Vals-1.1Legal4.27.5+3.3
Vals Legal IndexIndexes70.367.1−3.2
Design Arena Overall Code CategoriesCode12801297+2.8
LiveBench 2.0Agents75.673.1−2.5
Excel Modeling Benchmark v1.1Finance & Accounting65.567.8+2.4
APEX-Accounting 1.1Finance & Accounting62.260.0−2.2
Code Migration v1.1 (Vals Index v2.2 subset)Code50.248.2−2.0
TaxCalcBench TY25Finance & Accounting2.00.0−2.0
Vals Finance IndexIndexes57.455.9−1.5
Toloka Arena (v3)Agents21.022.0+1.0
GDPxProfessional Work4.94.0−0.9
Tax Agent BenchFinance & Accounting62.262.7+0.5
APEX-1 (Mercor, mean score)Professional Work64.765.0+0.3
APEX-Agents 1.1Professional Work68.368.1−0.2
NoCoT-Bench NCRI 15.2General QA68.368.40.0

Claude Haiku 5.5vsMiniMax M3.1 Flashnormalised score marginsData Nov 15, 2026Data: Vals AI, Mercor, Artificial Analysis and 13 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…