Skip to main content
BenchLM
Advertise

Model comparison

DeepSeek V4 Pro (Max) vs Kimi K3

Data verified

Head-to-head evidence from 27 shared benchmark results across 5 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

N/A
No comparison
Kimi K3
Moonshot AI
80.96/100
0 category wins2 category wins

Public leaderboard positions: DeepSeek V4 Pro (Max) unranked (Not scored); Kimi K3 #4 (Supported). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.

Evidence parity. DeepSeek V4 Pro (Max) and Kimi K3 share 27 comparable benchmark results. 2 of 8 categories are comparable. 21 results are unique to DeepSeek V4 Pro (Max); 30 to Kimi K3.

Updated July 23, 2026
Shared results
27
DeepSeek V4 Pro (Max) only
21
Kimi K3 only
30
Comparable categories
2 / 8

Treat this as a split decision. DeepSeek V4 Pro (Max) makes more sense if you want the cheaper token bill; Kimi K3 is the better fit if agentic is the priority or you need the larger 1.05M context window.

Confidence note. This is a partial-evidence comparison with 27 shared benchmark results across 5 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

DeepSeek V4 Pro (Max) and Kimi K3 finish on the same BenchAlign overall score, so this is less about a single winner and more about where the edge shows up. The BenchAlign headline says tie; the benchmark table is where the real choice happens.

Kimi K3 is also the more expensive model on tokens at $3.00 input / $15.00 output per 1M tokens, versus $0.43 input / $0.87 output per 1M tokens for DeepSeek V4 Pro (Max). That is roughly 17.2x on output cost alone. Kimi K3 gives you the larger context window at 1.05M, compared with 1M for DeepSeek V4 Pro (Max).

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 V4 Pro (Max) and Kimi K3
CategoryDeepSeek V4 Pro (Max)ΔKimi K3
AgenticDeepSeek V4 Pro (Max)74.5Margin→ 15.0Kimi K389.5
KnowledgeDeepSeek V4 Pro (Max)60.1Margin→ 0.9Kimi K361.0
CodingDeepSeek V4 Pro (Max)70.9MarginNo overlapKimi K3Not measured
MathDeepSeek V4 Pro (Max)95.2MarginNo overlapKimi K3Not measured
MultimodalDeepSeek V4 Pro (Max)Not measuredMarginNo overlapKimi K378.5

Decisive benchmark drivers

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

More
A · DeepSeek V4 Pro (Max)B · Kimi K3

  1. Terminal-Bench 2.0

    Agentic
    Source ↗
    A 67.9%B 88.3%
    Winner: Kimi K3Δ 20.4
    Terminal-Bench 2.0: DeepSeek V4 Pro (Max) scored 67.9%; Kimi K3 scored 88.3%. Kimi K3 wins this benchmark.

  2. HLE

    Knowledge
    Source ↗
    A 37.7%B 56%
    Winner: Kimi K3Δ 18.3
    HLE: DeepSeek V4 Pro (Max) scored 37.7%; Kimi K3 scored 56%. Kimi K3 wins this benchmark.

  3. BrowseComp

    Agentic
    Source ↗
    A 83.4%B 91.2%
    Winner: Kimi K3Δ 7.8
    BrowseComp: DeepSeek V4 Pro (Max) scored 83.4%; Kimi K3 scored 91.2%. Kimi K3 wins this benchmark.

  4. GPQA

    Knowledge
    Source ↗
    A 90.1%B 93.5%
    Winner: Kimi K3Δ 3.4
    GPQA: DeepSeek V4 Pro (Max) scored 90.1%; Kimi K3 scored 93.5%. Kimi K3 wins this benchmark.

Operational comparison

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

MetricDeepSeek V4 Pro (Max)Kimi K3Comparison
Input / output priceUSD per 1M tokensDeepSeek V4 Pro (Max)$0.435 input / $0.87 outputKimi K3$3 input / $15 outputDeepSeek V4 Pro (Max) has the lower combined listed price.
Generation speedtokens per secondDeepSeek V4 Pro (Max)Not availableKimi K3Not availableA complete speed comparison is not available.
First-answer latencyseconds to first tokenDeepSeek V4 Pro (Max)Not availableKimi K3Not availableA complete latency comparison is not available.
Context windowmaximum listed tokensDeepSeek V4 Pro (Max)1MKimi K31.05MKimi K3 lists the larger context window.

Benchmark Deep Dive

AgenticKimi K3 wins
BenchmarkDeepSeek V4 Pro (Max)Kimi K3Result
Terminal-Bench 2.0Source 67.9%88.3%Kimi K3 leads
BrowseCompSource 83.4%91.2%Kimi K3 leads
HLE w/ toolsSource 48.2%—Not comparable
MCP AtlasSource 73.6%84.2%Kimi K3 leads
GDPval-AASource 13071679Kimi K3 leads
ToolathlonSource 51.8%—Not comparable
AA Agentic IndexSource 36.4%50.1%Kimi K3 leads
APEX-Agents-AASource 24.3%41.3%Kimi K3 leads
τ²-bench resultsSource 96.2%—Not comparable
GDPval-AASource 40.4%59.0%Kimi K3 leads
AA BriefcaseSource 9321543Kimi K3 leads
AA EnterpriseOps-GymSource 40.4%45.3%Kimi K3 leads
AA Harvey LABSource 84.4%94.6%Kimi K3 leads
AA ITBenchSource 38.3%47.7%Kimi K3 leads
AA Tau3 BankingSource 25.8%33.4%Kimi K3 leads
terminalBenchHardSource 46.2%—Not comparable
aaTerminalBench21Source 64%85%Kimi K3 leads
DeepSearchQASource —95.0%Not comparable
Toolathlon-VerifiedSource —73.2%Not comparable
AutomationBenchSource —30.8%Not comparable
JobBenchSource —52.9%Not comparable
APEX-AgentsSource —37.6%Not comparable
SpreadsheetBench 2Source —34.8%Not comparable
DECK-BenchSource —73.5%Not comparable
AA AutomationBenchSource —52.7%Not comparable
Coding
BenchmarkDeepSeek V4 Pro (Max)Kimi K3Result
CodeforcesSource 3206.0—Not comparable
SWE-bench VerifiedSource 80.6%—Not comparable
SWE-bench ProSource 55.4%—Not comparable
SWE MultilingualSource 76.2%—Not comparable
Terminal-Bench 2.0Source 67.9%—Not comparable
Vibe Code BenchSource 49.93%—Not comparable
AA Coding IndexSource 59.4%76.2%Kimi K3 leads
AA-SciCodeSource 50.0%58.7%Kimi K3 leads
deepSweSource —67.5%Not comparable
FrontierSWESource —81.2%Not comparable
ProgramBenchSource —77.8%Not comparable
Kimi Code Bench v2Source —72.9%Not comparable
sweMarathonSource —42%Not comparable
PostTrain BenchSource —36.6%Not comparable
MLS-Bench LiteSource —48.3%Not comparable
Reasoning
BenchmarkDeepSeek V4 Pro (Max)Kimi K3Result
MRCR 1MSource 83.5%—Not comparable
CorpusQA 1MSource 62.0%—Not comparable
AA-LCRSource 66.3%74.7%Kimi K3 leads
CritPtSource 12.9%23.4%Kimi K3 leads
KnowledgeKimi K3 wins
BenchmarkDeepSeek V4 Pro (Max)Kimi K3Result
MMLU-ProSource 87.5%—Not comparable
SimpleQASource 57.9%—Not comparable
Chinese-SimpleQASource 84.4%—Not comparable
GPQASource 90.1%93.5%Kimi K3 leads
GPQA-DSource 90.1%93.5%Kimi K3 leads
HLESource 37.7%56%Kimi K3 leads
Artificial Analysis Intelligence IndexSource 44.3%57.1%Kimi K3 leads
AA-GPQA DiamondSource 88.8%93.5%Kimi K3 leads
AA-HLESource 35.9%44.3%Kimi K3 leads
AA-Omniscience IndexSource -10.0%18.4%Kimi K3 leads
AA-Omniscience AccuracySource 43.3%46.0%Kimi K3 leads
AA-Omniscience Hallucination RateSource 94.0%50.9%Kimi K3 leads
AA Openness IndexSource 50.0%—Not comparable
HLE w/o toolsSource —43.5%Not comparable
Math
BenchmarkDeepSeek V4 Pro (Max)Kimi K3Result
HMMT Feb 2026Source 95.2%—Not comparable
IMOAnswerBenchSource 89.8%—Not comparable
ApexSource 38.3%—Not comparable
Apex ShortlistSource 90.2%—Not comparable
Multimodal
BenchmarkDeepSeek V4 Pro (Max)Kimi K3Result
Design Arena WebsiteSource 12641386Kimi K3 leads
OfficeQA ProSource —63.3%Not comparable
MMMU-ProSource —81.6%Not comparable
MMMU-Pro w/ PythonSource —83.4%Not comparable
CharXiv w/o toolsSource —84.8%Not comparable
CharXivSource —91.3%Not comparable
MathVisionSource —94.3%Not comparable
MathVision w/ PythonSource —97.8%Not comparable
BabyVision w/ PythonSource —85.7%Not comparable
ZeroBenchSource —23.0%Not comparable
ZeroBench w/ PythonSource —41.0%Not comparable
WorldVQA ForceAnswerSource —51.0%Not comparable
OmniDocBenchSource —91.1%Not comparable
PerceptionBenchSource —58.5%Not comparable
AA-MMMU-ProSource —80.5%Not comparable
Inst. Following
BenchmarkDeepSeek V4 Pro (Max)Kimi K3Result
AA-IFBenchSource 76.5%—Not comparable
Frequently Asked Questions (3)

Which is better, DeepSeek V4 Pro (Max) or Kimi K3?

DeepSeek V4 Pro (Max) and Kimi K3 are tied on the BenchAlign overall score, so the right pick depends on which category matters most for your use case.

Which is better for knowledge tasks, DeepSeek V4 Pro (Max) or Kimi K3?

Kimi K3 has the edge for knowledge tasks in this comparison, averaging 61 versus 60.1. Inside this category, AA-Omniscience Hallucination Rate is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, DeepSeek V4 Pro (Max) or Kimi K3?

Kimi K3 has the edge for agentic tasks in this comparison, averaging 89.5 versus 74.5. Inside this category, AA Briefcase is the benchmark that creates the most daylight between them.

Related Comparisons

Explore More

DeepSeek Compare Pricing Methodology Find Your Best LLM Overall Rankings
Last updated: July 23, 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.