Skip to main content
BenchLM
Advertise

Model comparison

DeepSeek V3.2 vs Kimi K2.5

Data verified

Head-to-head evidence from 18 shared benchmark results across 7 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

55.4/100
Margin
4.3pts
winning →
Kimi K2.5
Moonshot AI
59.66/100
1 category wins1 category wins

Public leaderboard positions: DeepSeek V3.2 #82 (Supported); Kimi K2.5 #54 (Supported). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.

Evidence parity. DeepSeek V3.2 and Kimi K2.5 share 18 comparable benchmark results. 2 of 8 categories are comparable. 1 result is unique to DeepSeek V3.2; 45 to Kimi K2.5.

Updated July 20, 2026
Shared results
18
DeepSeek V3.2 only
1
Kimi K2.5 only
45
Comparable categories
2 / 8

Pick Kimi K2.5 if you want the stronger benchmark profile. DeepSeek V3.2 only becomes the better choice if coding is the priority or you want the cheaper token bill.

Confidence note. This is a partial-evidence comparison with 18 shared benchmark results across 7 evidence categories; 2 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.

Why this result

Kimi K2.5 is clearly ahead on the BenchAlign aggregate, 59.66 to 55.4. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.

Kimi K2.5's sharpest advantage is in mathematics, where it averages 60.6 against 17.1. The single biggest benchmark swing on the page is FrontierMath v2 (Tiers 1-3), 22.100% to 27.900%. DeepSeek V3.2 does hit back in coding, 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 $0.28 input / $0.42 output per 1M tokens for DeepSeek V3.2. That is roughly 7.1x on output cost alone. Kimi K2.5 gives you the larger context window at 256K, compared with 128K for DeepSeek V3.2.

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 DeepSeek V3.2 and Kimi K2.5
CategoryDeepSeek V3.2ΔKimi K2.5
MathDeepSeek V3.217.1Margin→ 43.5Kimi K2.560.6
CodingDeepSeek V3.260.9Margin← 1.5Kimi K2.559.4
AgenticDeepSeek V3.2Not measuredMarginNo overlapKimi K2.555.0
ReasoningDeepSeek V3.2Not measuredMarginNo overlapKimi K2.561.0
KnowledgeDeepSeek V3.2Not measuredMarginNo overlapKimi K2.556.9
MultilingualDeepSeek V3.2Not measuredMarginNo overlapKimi K2.582.3
MultimodalDeepSeek V3.2Not measuredMarginNo overlapKimi K2.578.5
Inst. FollowingDeepSeek V3.2Not measuredMarginNo overlapKimi K2.593.9

Decisive benchmark drivers

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

More
A · DeepSeek V3.2B · Kimi K2.5

  1. FrontierMath v2 (Tiers 1-3)

    Math
    Source ↗
    A 22.100%B 27.900%
    Winner: Kimi K2.5Δ 5.8
    FrontierMath v2 (Tiers 1-3): DeepSeek V3.2 scored 22.100%; Kimi K2.5 scored 27.900%. Kimi K2.5 wins this benchmark.

  2. SWE-Rebench

    Coding
    Source ↗
    A 60.9%B 58.5%
    Winner: DeepSeek V3.2Δ 2.4
    SWE-Rebench: DeepSeek V3.2 scored 60.9%; Kimi K2.5 scored 58.5%. DeepSeek V3.2 wins this benchmark.

  3. FrontierMath v2 (Tier 4)

    Math
    Source ↗
    A 2.100%B 4.200%
    Winner: Kimi K2.5Δ 2.1
    FrontierMath v2 (Tier 4): DeepSeek V3.2 scored 2.100%; Kimi K2.5 scored 4.200%. Kimi K2.5 wins this benchmark.

Operational comparison

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

MetricDeepSeek V3.2Kimi K2.5Comparison
Input / output priceUSD per 1M tokensDeepSeek V3.2$0.28 input / $0.42 outputKimi K2.5$0.6 input / $3 outputDeepSeek V3.2 has the lower combined listed price.
Generation speedtokens per secondDeepSeek V3.235 tok/sKimi K2.545 tok/sKimi K2.5 has the higher measured throughput.
First-answer latencyseconds to first tokenDeepSeek V3.23.75 sKimi K2.52.38 sKimi K2.5 reaches the first token sooner.
Context windowmaximum listed tokensDeepSeek V3.2128KKimi K2.5256KKimi K2.5 lists the larger context window.

Benchmark Deep Dive

Agentic
BenchmarkDeepSeek V3.2Kimi K2.5Result
Claw-EvalSource 40.2%52.3%Kimi K2.5 leads
VITA-BenchSource 18.5%—Not comparable
τ²-bench resultsSource 78.9%95.9%Kimi K2.5 leads
Gert LabsSource 29.57%45.88%Kimi K2.5 leads
Terminal-Bench 2.0Source —50.8%Not comparable
BrowseCompSource —60.6%Not comparable
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%Not comparable
MCP-TasksSource —59.1%Not comparable
WideResearchSource —72.7%Not comparable
APEX-Agents-AASource —11.5%Not comparable
ResearchClawBenchSource —14.0%Not comparable
JobBenchSource —8.7%Not comparable
AA Agentic IndexSource —21.7%Not comparable
GDPval-AASource —25.4%Not comparable
GDPval-AASource —1009Not comparable
CodingDeepSeek V3.2 wins
BenchmarkDeepSeek V3.2Kimi K2.5Result
SWE-RebenchSource 60.9%58.5%DeepSeek V3.2 leads
React Native EvalsSource 71.5%77.2%Kimi K2.5 leads
AA-SciCodeSource 38.7%49.0%Kimi K2.5 leads
SWE-bench VerifiedSource —76.8%Not comparable
SWE-bench Verified*Source —70.8%Not comparable
LiveCodeBench v6Source —85.0%Not comparable
SWE-bench ProSource —50.7%Not comparable
SWE MultilingualSource —73%Not comparable
SciCodeSource —48.7%Not comparable
AA Coding IndexSource —46.8%Not comparable
Reasoning
BenchmarkDeepSeek V3.2Kimi K2.5Result
AA-LCRSource 39.0%65.3%Kimi K2.5 leads
CritPtSource 0.9%3.1%Kimi K2.5 leads
LongBench v2Source —61%Not comparable
Knowledge
BenchmarkDeepSeek V3.2Kimi K2.5Result
Artificial Analysis Intelligence IndexSource 24.7%35.4%Kimi K2.5 leads
AA-GPQA DiamondSource 75.1%87.9%Kimi K2.5 leads
AA-HLESource 10.5%29.4%Kimi K2.5 leads
AA-Omniscience IndexSource -46.7%-8.1%Kimi K2.5 leads
AA-Omniscience AccuracySource 24.2%34.3%Kimi K2.5 leads
AA-Omniscience Hallucination RateSource 93.5%64.6%Kimi K2.5 leads
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
MathKimi K2.5 wins
BenchmarkDeepSeek V3.2Kimi K2.5Result
FrontierMath v2 (Tiers 1-3)Source 22.100%27.900%Kimi K2.5 leads
FrontierMath v2 (Tier 4)Source 2.100%4.200%Kimi K2.5 leads
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
Multilingual
BenchmarkDeepSeek V3.2Kimi K2.5Result
MMLU-ProXSource —82.3%Not comparable
NOVA-63Source —56.0%Not comparable
Multimodal
BenchmarkDeepSeek V3.2Kimi K2.5Result
Design Arena WebsiteSource 12061282Kimi K2.5 leads
MMMU-ProSource —78.5%Not comparable
Video-MMESource —87.4%Not comparable
MMVUSource —80.4%Not comparable
VideoMMMUSource —86.6%Not comparable
AA-MMMU-ProSource —75.4%Not comparable
Inst. Following
BenchmarkDeepSeek V3.2Kimi K2.5Result
AA-IFBenchSource 49.0%70.2%Kimi K2.5 leads
IFEvalSource —93.9%Not comparable
Frequently Asked Questions (3)

Which is better, DeepSeek V3.2 or Kimi K2.5?

Kimi K2.5 is ahead on BenchLM's BenchAlign leaderboard, 59.66 to 55.4. The biggest single separator in this matchup is FrontierMath v2 (Tiers 1-3), where the scores are 22.100% and 27.900%.

Which is better for coding, DeepSeek V3.2 or Kimi K2.5?

DeepSeek V3.2 has the edge for coding in this comparison, averaging 60.9 versus 59.4. Inside this category, AA-SciCode is the benchmark that creates the most daylight between them.

Which is better for math, DeepSeek V3.2 or Kimi K2.5?

Kimi K2.5 has the edge for math in this comparison, averaging 60.6 versus 17.1. Inside this category, FrontierMath v2 (Tiers 1-3) 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.

DeepSeek V3.2
API / mo$525
Self-host / moNot listed
Break-even—
Proprietary model — self-hosting not applicable.
Kimi K2.5
API / mo$2,700
Self-host / mo$5,221
Break-even132M/day
Model the full break-even

Related Comparisons

Explore More

DeepSeek Compare Pricing Methodology Find Your Best LLM Overall Rankings
Last updated: July 20, 2026

Choose a model with this week’s evidence

Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.

One email each week. Unsubscribe anytime.