BenchLeader

MiniMax M3.1 Flash vs Step 3.7 Flash

Full profiles: MiniMax M3.1 Flash (MiniMax) and Step 3.7 Flash (StepFun): every benchmark, price and rank.

Updated · Data from 9 sources

As of 16 Nov 2026, MiniMax M3.1 Flash and Step 3.7 Flash are neck and neck on quality (54.3 vs 53.0 on the BenchLeader Index, inside the ±10.1 margin), and MiniMax M3.1 Flash is 1.6× cheaper at $0.262 vs $0.416 per 1M tokens blended. MiniMax M3.1 Flash leads in every category we measure both in.

Key facts (as of 16 Nov 2026)

  • MiniMax M3.1 Flash scores 54.3 on the BenchLeader Index and Step 3.7 Flash scores 53.0 (BenchLeader, 16 Nov 2026).
  • MiniMax M3.1 Flash costs $0.262 per 1M tokens blended (list price, 16 Nov 2026).
  • Step 3.7 Flash costs $0.416 per 1M tokens blended (list price, 16 Nov 2026).
  • Of the 9 benchmarks both models have, MiniMax M3.1 Flash leads 8 (BenchLeader, 16 Nov 2026).

Which to pick for…

The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.

Use casePickWhyConfidence
CodingMiniMax M3.1 Flash55 vs 39 category score; SciCode (AA) 53.4% vs 43.9%High
Agents & tool useMiniMax M3.1 Flash55 vs 48 category score; GDPval-AA v2.1 1465 vs 864Medium
Long documentsMiniMax M3.1 Flash65 vs 61 category score; AA-LCR v1.1 79.7% vs 73.7%Medium
Images and other mediaMiniMax M3.1 Flash62 vs 58 category score; MMMU-Pro 79.1% vs 75.3%Medium
Lowest costMiniMax M3.1 Flash$0.262 vs $0.438 per 1M blended; agentic coding session $0.011 vs $0.017–
Fastest first answerStep 3.7 Flash14.3 s (Step 3.7 Flash) vs 43.3 s (MiniMax M3.1 Flash) to the first answer token–
High volume on a budgetNeither flagshipA smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs–

Key differences

  • MiniMax M3.1 Flash and Step 3.7 Flash are neck and neck on quality (54.3 vs 53.0, within the ±10.1 margin).
  • MiniMax M3.1 Flash is stronger in agents & tools, coding, long context, multimodal.
  • MiniMax M3.1 Flash is 1.6× cheaper ($0.262 vs $0.416 per 1M blended).

At a glance

SpecMiniMax M3.1 FlashStep 3.7 Flash
DeveloperMiniMaxStepFun
Released19 Oct 202629 May 2026
Context window (tokens)1,000,000256,000
Compared atMiniMax M3.1 FlashStep 3.7 Flash
Open weightsYesYes
Price in / out / cached input per 1M$0.150 / $0.600 / $0.015$0.200 / $1.15 / $0.040

Category by category

Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.

CategoryMiniMax M3.1 FlashStep 3.7 FlashLeader
Reasoning50.1–not comparable
Coding55.439.4MiniMax M3.1 Flash
Agents & tools55.248.0MiniMax M3.1 Flash
Knowledge45.9–not comparable
Instruction following–66.5not comparable
Multimodal61.557.6MiniMax M3.1 Flash
Long context64.561.3MiniMax M3.1 Flash

On one chart

Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.

2550 avg75CodingAgents & toolsLong contextMultimodalMiniMax M3.1 Flash, Coding: 55.4MiniMax M3.1 Flash, Agents & tools: 55.2MiniMax M3.1 Flash, Long context: 64.5MiniMax M3.1 Flash, Multimodal: 61.5Step 3.7 Flash, Coding: 39.4Step 3.7 Flash, Agents & tools: 48.0Step 3.7 Flash, Long context: 61.3Step 3.7 Flash, Multimodal: 57.6

Not drawn, measured for fewer than two of these models: Reasoning, Maths, Knowledge, Instruction following, Human preference.

Model, click to add or removeIndex

Where sources disagree

Across the 9 benchmarks both have, MiniMax M3.1 Flash leads 8, Step 3.7 Flash leads 1.

Overall indexMiniMax M3.1 FlashStep 3.7 FlashLeader
BenchLeader Index54.353.0neck and neck

Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).

Price, speed and latency

List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.

MeasureMiniMax M3.1 FlashStep 3.7 Flash
Blended price per 1M tokens$0.262$0.438
Input / output per 1M$0.150 / $0.600$0.200 / $1.15
Cached input per 1M$0.015$0.040
Price per task: chat reply$0.0002$0.0004
Price per task: summarize a 30-page report$0.0022$0.0031
Price per task: code edit$0.0018$0.0029
Price per task: agentic coding session$0.011$0.017
Price per task: structured extraction$0.0004$0.0006
Agentic coding session, input mostly cached$0.0053$0.0094
Cost per benchmark task (AA Index)$0.491–
Output tokens per task (AA Index)138k–
Of which reasoning103k–
Time per task (AA Index)15 min–
Output speed151 tok/s148 tok/s
Time to first answer43.3 s14.3 s
Time to a complete answer, last measured 13 Nov 2026141.5 s29.5 s

Each model is compared at its best-ranked reasoning-effort setting. MiniMax M3.1 Flash scores 54.3 and takes 43 s to its first answer token. Step 3.7 Flash scores 53.0 and takes 14 s to its first answer token. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.

Every benchmark

76 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.

MetricPublisherMiniMax M3.1 FlashStep 3.7 Flash
BenchLeader IndexBenchLeader54.353.0
Agents & tools scoreBenchLeader55.248.0
Coding scoreBenchLeader55.439.4
Knowledge scoreBenchLeader45.9–
Long context scoreBenchLeader64.561.3
Multimodal scoreBenchLeader61.557.6
Reasoning scoreBenchLeader50.1–
Instruction following scoreBenchLeader–66.5
Blended price $/MBenchLeader$0.262$0.438
Output speedBenchLeader151 tok/s148 tok/s
Time to first answerBenchLeader43.3 s14.3 s
Context windowBenchLeader1,000,000 tokens256,000 tokens
Reasoning
LiveBench 2.0 ReasoningLiveBench82.9%†–
GPQA Diamond (AA)Artificial Analysis–80.9%
Humanity's Last Exam (AA)Artificial Analysis48.0%21.4%
HLE-Diamond (AA)Artificial Analysis28.1%–
CritPtArtificial Analysis22.3%2.3%
CritPt v2 hard subset (AA)Artificial Analysis32.5%–
Coding
SciCodeSciCode–40.0%
Arena WebDevArena (LMArena)1584–
LiveBench 2.0 CodingLiveBench68.8%†–
SciCode (AA)Artificial Analysis53.4%43.9%
Vibe Code Bench v1.1Vals AI73.9%†–
Vibe Code Bench 1-100 v1.1Vals AI19.8%†–
SWE-Bench Pro V2Scale AI SEAL96.6%†–
SWE-Bench Pro V2 HARDScale AI SEAL84.3%†–
ALE-BenchEpoch AI Benchmarking Hub–694.1
Terminal-Bench 4.0 (AA)Artificial Analysis25.8%–
Terminal-Bench 2.1 (AA)Artificial Analysis–39.3%
GameBench 2 OverallDuelLab GameBench 2–23.2
GameBench 2 by reasoning settingDuelLab GameBench 230.3†29.4†
Design Arena WebsiteDesign Arena12771186
FrontierCode 1.1 (AA)Artificial Analysis38.6%–
FrontierCode 2.0 (AA)Artificial Analysis20.8%–
Agents & tools
Arena AgentArena (LMArena)-2.7–
LiveBench Long AgenticLiveBench25.7%†–
LiveBench BrowserLiveBench80.6%†–
Show all 76 benchmarks
MetricPublisherMiniMax M3.1 FlashStep 3.7 Flash
Agents & tools
Terminal-Bench HardArtificial Analysis–35.6%
τ²-Bench Telecom (AA)Artificial Analysis–98.5%
HiL-BenchScale AI SEAL65.7%–
GDPval-AA v2.1, normalised scoreArtificial Analysis47.9%–
GDPval-AA v2.1Artificial Analysis1465864
τ³-Banking (AA)Artificial Analysis–12.0%
ITBench SRE (AA)Artificial Analysis–30.3%
APEX-Agents (AA)Artificial Analysis–14.8%
Finance Agent v2.1Vals AI58.8%†–
Harvey's Legal Agent Benchmark v1.1Vals AI7.5%†–
Legal Research BenchVals AI34.1%†–
Tax Agent BenchVals AI32.1%†–
Terminal-Bench 4.0 (Vals)Vals AI25.3%†–
Arena Agent Task CompletionArena (LMArena)-0.7–
AutomationBenchArtificial Analysis72.5%–
GDP.pdf (AA)Artificial Analysis20.8%–
EnterpriseOps-GymArtificial Analysis–29.1%
AA-Briefcase v1.1Artificial Analysis1474–
GDPxGDPx by SemiAnalysis4.0%†–
Design Arena AgenticDesign Arena1252–
Terminal-Bench-Science 0.1 (AA)Artificial Analysis3.3%–
MKA (AA)Artificial Analysis33.5%–
Maths
AIME 2026MathArena–95.0%†
HMMT February 2026MathArena–87.9%†
MathArena ApexMathArena–14.6%†
Knowledge
LiveBench 2.0 Data AnalysisLiveBench71.4%†–
LiveBench ResearchLiveBench76.7%†–
AA-Omniscience v1.1Artificial Analysis-6.4–
PRBench FinanceScale AI SEAL45.6%–
PRBench LegalScale AI SEAL45.1%–
Excel Modeling Benchmark v1.1Vals AI67.8%†–
Access to Justice BenchmarkVals AI73.3%†–
AA-Omniscience v1.1: accuracyArtificial Analysis45.1%–
AA-Omniscience v1.1: non-hallucinationArtificial Analysis6.3%–
Instruction following
IFBenchArtificial Analysis–67.3%
Multimodal
LiveBench DesignLiveBench71.2%†–
MMMU-ProArtificial Analysis79.1%75.3%
SAGEVals AI85.3%†–
VIALS-AAHandshake AI29.8%–
Long context
AA-LCR v1.1Artificial Analysis79.7%73.7%
MLCRArtificial Analysis16.7%–
Composite
LiveBench 2.0LiveBench63.7%†–
AA Intelligence Index v5.0.1Artificial Analysis29.1–
AA Intelligence Index v4.3.2Artificial Analysis42.719.5
AA Progress Index v1Artificial Analysis317.6264.2
Vals IndexVals AI53.8†–
Vals Finance IndexVals AI55.9%†–
Vals Router IndexVals AI53.8%–
Vals Legal IndexVals AI67.1%†–
Safety & honesty
FORTRESSScale AI SEAL52.1%–
Metadata
AA Openness IndexArtificial Analysis27.838.9

Data as of 2026-11-16. Best configuration of each model; every score links to its source on the model pages.

DuelLab GameBench 2, release gb2-20261111-7b598bb13de4 (https://benchmarks.duellab.org/releases/gb2-20261111-7b598bb13de4/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.

What changed

  • MiniMax M3.1 Flash: BenchLeader Index 53.7 → 54.3.
  • Step 3.7 Flash: 1 new result (GDPval-AA v2.1).
  • Step 3.7 Flash: 1 score revision.
  • MiniMax M3.1 Flash: BenchLeader Index 54.3 → 53.7.
  • MiniMax M3.1 Flash: 1 score revision.
  • Step 3.7 Flash: 1 score revision.
  • MiniMax M3.1 Flash: 6 score revisions.
  • MiniMax M3.1 Flash: BenchLeader Index 53.0 → 54.3.
  • MiniMax M3.1 Flash: 1 new result (HiL-Bench).
  • MiniMax M3.1 Flash: BenchLeader Index 52.0 → 53.0.

Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs

MiniMax M3.1 Flash vs Step 3.7 Flash: questions

Is MiniMax M3.1 Flash better than Step 3.7 Flash?
MiniMax M3.1 Flash and Step 3.7 Flash are neck and neck on quality (54.3 vs 53.0, within the ±10.1 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 16 Nov 2026, so check the category scores for your use.
MiniMax M3.1 Flash vs Step 3.7 Flash: which is cheaper, and will that change?
MiniMax M3.1 Flash is cheaper today: $0.262 against $0.416 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
Is MiniMax M3.1 Flash or Step 3.7 Flash better for coding?
MiniMax M3.1 Flash scores higher in coding (55 vs 39 on the category index, where 50 is average).
Which is faster, MiniMax M3.1 Flash or Step 3.7 Flash?
MiniMax M3.1 Flash streams faster (151 vs 148 output tokens per second, Artificial Analysis 7-day median).
Is MiniMax M3.1 Flash or Step 3.7 Flash better for agents and tool use?
MiniMax M3.1 Flash scores higher in agentic tasks (55 vs 48 on the category index, where 50 is average).
More questions (4)
Which is better for long documents?
MiniMax M3.1 Flash scores higher on long-context benchmarks (65 vs 61). MiniMax M3.1 Flash accepts more context: 1,000,000 against 256,000 tokens for Step 3.7 Flash.
Which has lower latency?
Step 3.7 Flash answers first: 14.3 s against 43.3 s to the first answer token, as Artificial Analysis measures it. Reasoning effort drives most of this: a higher setting thinks longer before it answers.
Do independent benchmarks agree on MiniMax M3.1 Flash vs Step 3.7 Flash?
Across the 9 benchmarks both have, MiniMax M3.1 Flash leads 8, Step 3.7 Flash leads 1.
When was MiniMax M3.1 Flash released?
MiniMax M3.1 Flash was released on 19 Oct 2026.