Skip to main content
BenchLM
Advertise

Model comparison

Claude Sonnet 4.6 vs Kimi K2.5

Data verified

Head-to-head evidence from 26 shared benchmark results across 7 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

65.07/100
Margin
5.4pts
← winning
Kimi K2.5
Moonshot AI
59.66/100
3 category wins2 category wins

Public leaderboard positions: Claude Sonnet 4.6 #32 (Supported); Kimi K2.5 #54 (Supported). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.

Evidence parity. Claude Sonnet 4.6 and Kimi K2.5 share 26 comparable benchmark results. 5 of 8 categories are comparable. 7 results are unique to Claude Sonnet 4.6; 37 to Kimi K2.5.

Updated July 20, 2026
Shared results
26
Claude Sonnet 4.6 only
7
Kimi K2.5 only
37
Comparable categories
5 / 8

Pick Claude Sonnet 4.6 if you want the stronger benchmark profile. Kimi K2.5 only becomes the better choice if mathematics is the priority or you want the cheaper token bill.

Confidence note. This is a partial-evidence comparison with 26 shared benchmark results across 7 evidence categories; 5 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.

Why this result

Claude Sonnet 4.6 is clearly ahead on the BenchAlign aggregate, 65.07 to 59.66. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.

Claude Sonnet 4.6's sharpest advantage is in agentic, where it averages 65.2 against 55. The single biggest benchmark swing on the page is SuperGPQA, 95% to 69.2%. Kimi K2.5 does hit back in mathematics, so the answer changes if that is the part of the workload you care about most.

Claude Sonnet 4.6 is also the more expensive model on tokens at $3.00 input / $15.00 output per 1M tokens, versus $0.60 input / $3.00 output per 1M tokens for Kimi K2.5. That is roughly 5.0x on output cost alone. Kimi K2.5 gives you the larger context window at 256K, compared with 200K for Claude Sonnet 4.6.

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 Claude Sonnet 4.6 and Kimi K2.5
CategoryClaude Sonnet 4.6ΔKimi K2.5
MathClaude Sonnet 4.626.4Margin→ 34.2Kimi K2.560.6
AgenticClaude Sonnet 4.665.2Margin← 10.2Kimi K2.555.0
CodingClaude Sonnet 4.669.1Margin← 9.7Kimi K2.559.4
KnowledgeClaude Sonnet 4.666.0Margin← 9.1Kimi K2.556.9
MultimodalClaude Sonnet 4.677.4Margin→ 1.1Kimi K2.578.5
ReasoningClaude Sonnet 4.6Not measuredMarginNo overlapKimi K2.561.0
MultilingualClaude Sonnet 4.6Not measuredMarginNo overlapKimi K2.582.3
Inst. FollowingClaude Sonnet 4.6Not measuredMarginNo overlapKimi K2.593.9

Decisive benchmark drivers

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

More
A · Claude Sonnet 4.6B · Kimi K2.5

  1. SuperGPQA

    Knowledge
    Source ↗
    A 95%B 69.2%
    Winner: Claude Sonnet 4.6Δ 25.8
    SuperGPQA: Claude Sonnet 4.6 scored 95%; Kimi K2.5 scored 69.2%. Claude Sonnet 4.6 wins this benchmark.

  2. HLE

    Knowledge
    Source ↗
    A 49%B 30.1%
    Winner: Claude Sonnet 4.6Δ 18.9
    HLE: Claude Sonnet 4.6 scored 49%; Kimi K2.5 scored 30.1%. Claude Sonnet 4.6 wins this benchmark.

  3. Terminal-Bench 2.0

    Agentic
    Source ↗
    A 59.1%B 50.8%
    Winner: Claude Sonnet 4.6Δ 8.3
    Terminal-Bench 2.0: Claude Sonnet 4.6 scored 59.1%; Kimi K2.5 scored 50.8%. Claude Sonnet 4.6 wins this benchmark.

  4. MMLU-Pro

    Knowledge
    Source ↗
    A 79.2%B 87.1%
    Winner: Kimi K2.5Δ 7.9
    MMLU-Pro: Claude Sonnet 4.6 scored 79.2%; Kimi K2.5 scored 87.1%. Kimi K2.5 wins this benchmark.

  5. FrontierMath v2 (Tiers 1-3)

    Math
    Source ↗
    A 32.400%B 27.900%
    Winner: Claude Sonnet 4.6Δ 4.5
    FrontierMath v2 (Tiers 1-3): Claude Sonnet 4.6 scored 32.400%; Kimi K2.5 scored 27.900%. Claude Sonnet 4.6 wins this benchmark.

Operational comparison

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

MetricClaude Sonnet 4.6Kimi K2.5Comparison
Input / output priceUSD per 1M tokensClaude Sonnet 4.6$3 input / $15 outputKimi K2.5$0.6 input / $3 outputKimi K2.5 has the lower combined listed price.
Generation speedtokens per secondClaude Sonnet 4.644 tok/sKimi K2.545 tok/sKimi K2.5 has the higher measured throughput.
First-answer latencyseconds to first tokenClaude Sonnet 4.61.48 sKimi K2.52.38 sClaude Sonnet 4.6 reaches the first token sooner.
Context windowmaximum listed tokensClaude Sonnet 4.6200KKimi K2.5256KKimi K2.5 lists the larger context window.

Benchmark Deep Dive

AgenticClaude Sonnet 4.6 wins
BenchmarkClaude Sonnet 4.6Kimi K2.5Result
Terminal-Bench 2.0Source 59.1%50.8%Claude Sonnet 4.6 leads
OSWorld-VerifiedSource 72.1%—Not comparable
Claw-EvalSource 67.8%52.3%Claude Sonnet 4.6 leads
CyberGymSource 65.2%—Not comparable
τ²-bench resultsSource 79.5%95.9%Kimi K2.5 leads
Gert LabsSource 62.92%45.88%Claude Sonnet 4.6 leads
OSWorld 2.0Source 8.3%—Not comparable
JobBenchSource 36.9%8.7%Claude Sonnet 4.6 leads
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
AA Agentic IndexSource —21.7%Not comparable
GDPval-AASource —25.4%Not comparable
GDPval-AASource —1009Not comparable
CodingClaude Sonnet 4.6 wins
BenchmarkClaude Sonnet 4.6Kimi K2.5Result
SWE-bench VerifiedSource 79.6%76.8%Claude Sonnet 4.6 leads
SWE-RebenchSource 60.7%58.5%Claude Sonnet 4.6 leads
React Native EvalsSource 80.6%77.2%Claude Sonnet 4.6 leads
Vibe Code BenchSource 51.48%—Not comparable
cursorBench31Source 48.8%—Not comparable
AA-SciCodeSource 46.9%49.0%Kimi K2.5 leads
FrontierCode 1.1 MainSource 24.3%—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
BenchmarkClaude Sonnet 4.6Kimi K2.5Result
AA-LCRSource 57.7%65.3%Kimi K2.5 leads
CritPtSource 0.9%3.1%Kimi K2.5 leads
LongBench v2Source —61%Not comparable
KnowledgeClaude Sonnet 4.6 wins
BenchmarkClaude Sonnet 4.6Kimi K2.5Result
GPQASource 89.9%87.6%Claude Sonnet 4.6 leads
SuperGPQASource 95%69.2%Claude Sonnet 4.6 leads
MMLU-ProSource 79.2%87.1%Kimi K2.5 leads
HLESource 49%30.1%Claude Sonnet 4.6 leads
Artificial Analysis Intelligence IndexSource 35.9%35.4%Claude Sonnet 4.6 leads
AA-GPQA DiamondSource 79.9%87.9%Kimi K2.5 leads
AA-HLESource 13.2%29.4%Kimi K2.5 leads
AA-Omniscience IndexSource -2.9%-8.1%Claude Sonnet 4.6 leads
AA-Omniscience AccuracySource 38.0%34.3%Claude Sonnet 4.6 leads
AA-Omniscience Hallucination RateSource 65.9%64.6%Kimi K2.5 leads
GPQA-DSource —87.6%Not comparable
MMLU-Pro (Arcee)Source —87.1%Not comparable
MathKimi K2.5 wins
BenchmarkClaude Sonnet 4.6Kimi K2.5Result
FrontierMath v2 (Tiers 1-3)Source 32.400%27.900%Claude Sonnet 4.6 leads
FrontierMath v2 (Tier 4)Source 8.300%4.200%Claude Sonnet 4.6 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
BenchmarkClaude Sonnet 4.6Kimi K2.5Result
MMLU-ProXSource —82.3%Not comparable
NOVA-63Source —56.0%Not comparable
MultimodalKimi K2.5 wins
BenchmarkClaude Sonnet 4.6Kimi K2.5Result
CharXivSource 77.4%—Not comparable
AA-MMMU-ProSource 70.6%75.4%Kimi K2.5 leads
Design Arena WebsiteSource 13171282Claude Sonnet 4.6 leads
MMMU-ProSource —78.5%Not comparable
Video-MMESource —87.4%Not comparable
MMVUSource —80.4%Not comparable
VideoMMMUSource —86.6%Not comparable
Inst. Following
BenchmarkClaude Sonnet 4.6Kimi K2.5Result
AA-IFBenchSource 41.2%70.2%Kimi K2.5 leads
IFEvalSource —93.9%Not comparable
Frequently Asked Questions (6)

Which is better, Claude Sonnet 4.6 or Kimi K2.5?

Claude Sonnet 4.6 is ahead on BenchLM's BenchAlign leaderboard, 65.07 to 59.66. The biggest single separator in this matchup is SuperGPQA, where the scores are 95% and 69.2%.

Which is better for knowledge tasks, Claude Sonnet 4.6 or Kimi K2.5?

Claude Sonnet 4.6 has the edge for knowledge tasks in this comparison, averaging 66 versus 56.9. Inside this category, SuperGPQA is the benchmark that creates the most daylight between them.

Which is better for coding, Claude Sonnet 4.6 or Kimi K2.5?

Claude Sonnet 4.6 has the edge for coding in this comparison, averaging 69.1 versus 59.4. Inside this category, React Native Evals is the benchmark that creates the most daylight between them.

Which is better for math, Claude Sonnet 4.6 or Kimi K2.5?

Kimi K2.5 has the edge for math in this comparison, averaging 60.6 versus 26.4. Inside this category, FrontierMath v2 (Tiers 1-3) is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, Claude Sonnet 4.6 or Kimi K2.5?

Claude Sonnet 4.6 has the edge for agentic tasks in this comparison, averaging 65.2 versus 55. Inside this category, JobBench is the benchmark that creates the most daylight between them.

Which is better for multimodal and grounded tasks, Claude Sonnet 4.6 or Kimi K2.5?

Kimi K2.5 has the edge for multimodal and grounded tasks in this comparison, averaging 78.5 versus 77.4. 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.

Claude Sonnet 4.6
API / mo$13,500
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

Anthropic 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.