Skip to main content
BenchLM
Advertise

Model comparison

Kimi K2.5 vs MiniMax M3

Data verified

Head-to-head evidence from 27 shared benchmark results across 6 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

Kimi K2.5
Moonshot AI
59.58/100
Margin
10.2pts
winning →
MiniMax M3
MiniMax
69.8/100
1 category wins3 category wins

Verified leaderboard positions: Kimi K2.5 #22; MiniMax M3 #18

BenchAlign evidence: Kimi K2.5 supported; MiniMax M3 supported. Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.

Evidence parity. Kimi K2.5 and MiniMax M3 share 27 comparable benchmark results. 4 of 8 categories are comparable. 37 results are unique to Kimi K2.5; 18 to MiniMax M3.

Updated July 16, 2026
Shared results
27
Kimi K2.5 only
37
MiniMax M3 only
18
Comparable categories
4 / 8

Pick MiniMax M3 if you want the stronger benchmark profile. Kimi K2.5 only becomes the better choice if multimodal & grounded is the priority.

Confidence note. This is a partial-evidence comparison with 27 shared benchmark results across 6 evidence categories; 4 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.

Why this result

MiniMax M3 is clearly ahead on the provisional aggregate, 70 to 61. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.

MiniMax M3's sharpest advantage is in mathematics, where it averages 85.7 against 60.6. The single biggest benchmark swing on the page is BrowseComp, 60.6% to 83.5%. Kimi K2.5 does hit back in multimodal & grounded, so the answer changes if that is the part of the workload you care about most.

Kimi K2.5 is also the more expensive model on tokens at $0.60 input / $3.00 output per 1M tokens, versus $0.30 input / $1.20 output per 1M tokens for MiniMax M3. That is roughly 2.5x on output cost alone. MiniMax M3 gives you the larger context window at 1M, compared with 256K for Kimi K2.5.

Category breakdown

Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.

Category scores and score margins for Kimi K2.5 and MiniMax M3
CategoryKimi K2.5ΔMiniMax M3
MathKimi K2.560.6Margin→ 25.1MiniMax M385.7
AgenticKimi K2.555.0Margin→ 17.3MiniMax M372.3
MultimodalKimi K2.578.5Margin← 13.6MiniMax M364.9
CodingKimi K2.559.4Margin→ 12.8MiniMax M372.2
ReasoningKimi K2.561.0MarginNo overlapMiniMax M3Not measured
KnowledgeKimi K2.557.2MarginNo overlapMiniMax M3Not measured
MultilingualKimi K2.582.3MarginNo overlapMiniMax M3Not measured
Inst. FollowingKimi K2.593.9MarginNo overlapMiniMax M3Not measured

Decisive benchmark drivers

The largest measured benchmark gaps in this matchup, with exact reported values.

More
A · Kimi K2.5B · MiniMax M3

  1. BrowseComp

    Agentic
    Source ↗
    A 60.6%B 83.5%
    Winner: MiniMax M3Δ 22.9
    BrowseComp: Kimi K2.5 scored 60.6%; MiniMax M3 scored 83.5%. MiniMax M3 wins this benchmark.

  2. Terminal-Bench 2.0

    Agentic
    Source ↗
    A 50.8%B 66%
    Winner: MiniMax M3Δ 15.2
    Terminal-Bench 2.0: Kimi K2.5 scored 50.8%; MiniMax M3 scored 66%. MiniMax M3 wins this benchmark.

  3. SWE-bench Pro

    Coding
    Source ↗
    A 50.7%B 59%
    Winner: MiniMax M3Δ 8.3
    SWE-bench Pro: Kimi K2.5 scored 50.7%; MiniMax M3 scored 59%. MiniMax M3 wins this benchmark.

  4. SWE-bench Verified

    Coding
    Source ↗
    A 76.8%B 80.5%
    Winner: MiniMax M3Δ 3.7
    SWE-bench Verified: Kimi K2.5 scored 76.8%; MiniMax M3 scored 80.5%. MiniMax M3 wins this benchmark.

  5. MMMU-Pro

    Multimodal
    Source ↗
    A 78.5%B 78.1%
    Winner: Kimi K2.5Δ 0.4
    MMMU-Pro: Kimi K2.5 scored 78.5%; MiniMax M3 scored 78.1%. Kimi K2.5 wins this benchmark.

Operational comparison

Runtime and commercial metrics are compared only when both models have a complete sourced value.

MetricKimi K2.5MiniMax M3Comparison
Input / output priceUSD per 1M tokensKimi K2.5$0.6 input / $3 outputMiniMax M3$0.3 input / $1.2 outputMiniMax M3 has the lower combined listed price.
Generation speedtokens per secondKimi K2.545 tok/sMiniMax M3Not availableA complete speed comparison is not available.
First-answer latencyseconds to first tokenKimi K2.52.38 sMiniMax M3Not availableA complete latency comparison is not available.
Context windowmaximum listed tokensKimi K2.5256KMiniMax M31MMiniMax M3 lists the larger context window.

Benchmark Deep Dive

AgenticMiniMax M3 wins
BenchmarkKimi K2.5MiniMax M3Result
Terminal-Bench 2.0Source 50.8%66%MiniMax M3 leads
BrowseCompSource 60.6%83.5%MiniMax M3 leads
Claw-EvalSource 52.3%74.5%MiniMax M3 leads
QwenClawBenchSource 54.3%—Not comparable
τ³-bench resultsSource 65.7%—Not comparable
DeepSearchQASource 77.1%—Not comparable
DeepPlanningSource 14.4%—Not comparable
ToolathlonSource 27.8%—Not comparable
MCP AtlasSource 29.5%74.2%MiniMax M3 leads
MCP-TasksSource 59.1%—Not comparable
WideResearchSource 72.7%—Not comparable
τ²-bench resultsSource 95.9%88.9%Kimi K2.5 leads
APEX-Agents-AASource 11.5%—Not comparable
Gert LabsSource 45.88%—Not comparable
ResearchClawBenchSource 14.0%19.8%MiniMax M3 leads
JobBenchSource 8.7%—Not comparable
AA Agentic IndexSource 21.7%35.4%MiniMax M3 leads
GDPval-AASource 25.4%44.7%MiniMax M3 leads
GDPval-AASource 10091395MiniMax M3 leads
OSWorld-VerifiedSource —70.1%Not comparable
GDPval rubricsSource —74.7%Not comparable
BankerToolBenchSource —76.1%Not comparable
OSWorld 2.0Source —4.6%Not comparable
AA BriefcaseSource —1110Not comparable
AA EnterpriseOps-GymSource —32.1%Not comparable
AA Harvey LABSource —6.7%Not comparable
CodingMiniMax M3 wins
BenchmarkKimi K2.5MiniMax M3Result
SWE-bench VerifiedSource 76.8%80.5%MiniMax M3 leads
SWE-bench Verified*Source 70.8%—Not comparable
LiveCodeBench v6Source 85.0%—Not comparable
SWE-bench ProSource 50.7%59%MiniMax M3 leads
SWE MultilingualSource 73%—Not comparable
SWE-RebenchSource 58.5%—Not comparable
React Native EvalsSource 77.2%—Not comparable
SciCodeSource 48.7%—Not comparable
Terminal-Bench HardSource 34.8%42.4%MiniMax M3 leads
AA-SciCodeSource 49.0%45.4%Kimi K2.5 leads
AA Coding IndexSource 46.8%58.6%MiniMax M3 leads
Terminal-Bench 2.0Source —66.0%Not comparable
NL2RepoSource —42.1%Not comparable
VIBE V2Source —50.1%Not comparable
SVG-BenchSource —63.7%Not comparable
KernelBench HardSource —28.8%Not comparable
AA Terminal-Bench 2.1Source —65.2%Not comparable
Reasoning
BenchmarkKimi K2.5MiniMax M3Result
LongBench v2Source 61%—Not comparable
AA-LCRSource 65.3%74.0%MiniMax M3 leads
CritPtSource 3.1%3.7%MiniMax M3 leads
Knowledge
BenchmarkKimi K2.5MiniMax M3Result
GPQASource 87.6%—Not comparable
GPQA-DSource 87.6%—Not comparable
SuperGPQASource 69.2%—Not comparable
MMLU-ProSource 87.1%—Not comparable
MMLU-Pro (Arcee)Source 87.1%—Not comparable
HLESource 30.1%—Not comparable
Artificial Analysis Intelligence IndexSource 35.4%44.4%MiniMax M3 leads
AA-GPQA DiamondSource 87.9%92.9%MiniMax M3 leads
AA-HLESource 29.4%37.1%MiniMax M3 leads
AA-Omniscience IndexSource -8.1%1.4%MiniMax M3 leads
AA-Omniscience AccuracySource 34.3%15.0%Kimi K2.5 leads
AA-Omniscience Hallucination RateSource 64.6%16.1%MiniMax M3 leads
AA Openness IndexSource —33.3%Not comparable
MathMiniMax M3 wins
BenchmarkKimi K2.5MiniMax M3Result
AIME 2025Source 96.1%—Not comparable
AIME26Source 95.8%—Not comparable
AIME25 (Arcee)Source 96.3%—Not comparable
HMMT Feb 2025Source 95.4%—Not comparable
HMMT Nov 2025Source 91.1%—Not comparable
HMMT Feb 2026Source 87.1%—Not comparable
MMAnswerBenchSource 81.8%—Not comparable
FrontierMath v2 (Tiers 1-3)Source 27.900%—Not comparable
FrontierMath v2 (Tier 4)Source 4.200%—Not comparable
USAMO 2026Source —85.7%Not comparable
Multilingual
BenchmarkKimi K2.5MiniMax M3Result
MMLU-ProXSource 82.3%—Not comparable
NOVA-63Source 56.0%—Not comparable
MultimodalKimi K2.5 wins
BenchmarkKimi K2.5MiniMax M3Result
MMMU-ProSource 78.5%78.1%Kimi K2.5 leads
Video-MMESource 87.4%—Not comparable
MMVUSource 80.4%—Not comparable
VideoMMMUSource 86.6%84.6%Kimi K2.5 leads
AA-MMMU-ProSource 75.4%78.6%MiniMax M3 leads
Design Arena WebsiteSource 12841294MiniMax M3 leads
OfficeQA ProSource —45.1%Not comparable
OmniDocBench 1.5Source —91.6%Not comparable
Video-MME (with subtitle)Source —85.4%Not comparable
Inst. Following
BenchmarkKimi K2.5MiniMax M3Result
IFEvalSource 93.9%—Not comparable
AA-IFBenchSource 70.2%82.9%MiniMax M3 leads
Frequently Asked Questions (5)

Which is better, Kimi K2.5 or MiniMax M3?

MiniMax M3 is ahead on BenchLM's provisional leaderboard, 70 to 61. The biggest single separator in this matchup is BrowseComp, where the scores are 60.6% and 83.5%.

Which is better for coding, Kimi K2.5 or MiniMax M3?

MiniMax M3 has the edge for coding in this comparison, averaging 72.2 versus 59.4. Inside this category, AA Coding Index is the benchmark that creates the most daylight between them.

Which is better for math, Kimi K2.5 or MiniMax M3?

MiniMax M3 has the edge for math in this comparison, averaging 85.7 versus 60.6. Kimi K2.5 stays close enough that the answer can still flip depending on your workload.

Which is better for agentic tasks, Kimi K2.5 or MiniMax M3?

MiniMax M3 has the edge for agentic tasks in this comparison, averaging 72.3 versus 55. Inside this category, GDPval-AA is the benchmark that creates the most daylight between them.

Which is better for multimodal and grounded tasks, Kimi K2.5 or MiniMax M3?

Kimi K2.5 has the edge for multimodal and grounded tasks in this comparison, averaging 78.5 versus 64.9. Inside this category, Design Arena Website is the benchmark that creates the most daylight between them.

Self-host vs API cost

Estimates at 50,000 req/day · 1000 tokens/req average.

Kimi K2.5
API / mo$2,700
Self-host / mo$5,221
Break-even132M/day
MiniMax M3
API / mo$1,125
Self-host / moNot listed
Break-even—
Proprietary model — self-hosting not applicable.
Model the full break-even

Related Comparisons

Explore More

Moonshot AI Compare Pricing Methodology Find Your Best LLM Overall Rankings
Last updated: July 16, 2026

The AI models change fast. We track them for you.

A weekly brief for engineers and researchers covering new models, ranking shifts, and pricing changes.

Free. No spam. Unsubscribe anytime.