Skip to main content
BenchLM
Advertise

Model comparison

Claude Opus 4.7 vs Kimi K2.5

Data verified

Head-to-head evidence from 19 shared benchmark results across 7 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

71.63/100
Margin
12.0pts
← winning
Kimi K2.5
Moonshot AI
59.58/100
0 category wins1 category wins

Verified leaderboard positions: Claude Opus 4.7 unranked; Kimi K2.5 #22

BenchAlign evidence: Claude Opus 4.7 supported; Kimi K2.5 supported. Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.

Evidence parity. Claude Opus 4.7 and Kimi K2.5 share 19 comparable benchmark results. 1 of 8 categories are comparable. 3 results are unique to Claude Opus 4.7; 45 to Kimi K2.5.

Updated July 15, 2026
Shared results
19
Claude Opus 4.7 only
3
Kimi K2.5 only
45
Comparable categories
1 / 8

Pick Claude Opus 4.7 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 19 shared benchmark results across 7 evidence categories; 1 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.

Why this result

Claude Opus 4.7 is clearly ahead on the provisional aggregate, 69 to 61. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.

Claude Opus 4.7 is also the more expensive model on tokens at $5.00 input / $25.00 output per 1M tokens, versus $0.60 input / $3.00 output per 1M tokens for Kimi K2.5. That is roughly 8.3x on output cost alone. Claude Opus 4.7 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 Claude Opus 4.7 and Kimi K2.5
CategoryClaude Opus 4.7ΔKimi K2.5
MathClaude Opus 4.738.6Margin→ 22.0Kimi K2.560.6
AgenticClaude Opus 4.7Not measuredMarginNo overlapKimi K2.555.0
CodingClaude Opus 4.7Not measuredMarginNo overlapKimi K2.559.4
ReasoningClaude Opus 4.7Not measuredMarginNo overlapKimi K2.561.0
KnowledgeClaude Opus 4.7Not measuredMarginNo overlapKimi K2.557.2
MultilingualClaude Opus 4.7Not measuredMarginNo overlapKimi K2.582.3
MultimodalClaude Opus 4.7Not measuredMarginNo overlapKimi K2.578.5
Inst. FollowingClaude Opus 4.7Not measuredMarginNo overlapKimi K2.593.9

Decisive benchmark drivers

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

More
A · Claude Opus 4.7B · Kimi K2.5

  1. FrontierMath v2 (Tier 4)

    Math
    Source ↗
    A 22.917%B 4.200%
    Winner: Claude Opus 4.7Δ 18.7
    FrontierMath v2 (Tier 4): Claude Opus 4.7 scored 22.917%; Kimi K2.5 scored 4.200%. Claude Opus 4.7 wins this benchmark.

  2. FrontierMath v2 (Tiers 1-3)

    Math
    Source ↗
    A 43.793%B 27.900%
    Winner: Claude Opus 4.7Δ 15.9
    FrontierMath v2 (Tiers 1-3): Claude Opus 4.7 scored 43.793%; Kimi K2.5 scored 27.900%. Claude Opus 4.7 wins this benchmark.

Operational comparison

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

MetricClaude Opus 4.7Kimi K2.5Comparison
Input / output priceUSD per 1M tokensClaude Opus 4.7$5 input / $25 outputKimi K2.5$0.6 input / $3 outputKimi K2.5 has the lower combined listed price.
Generation speedtokens per secondClaude Opus 4.7Not availableKimi K2.545 tok/sA complete speed comparison is not available.
First-answer latencyseconds to first tokenClaude Opus 4.7Not availableKimi K2.52.38 sA complete latency comparison is not available.
Context windowmaximum listed tokensClaude Opus 4.71MKimi K2.5256KClaude Opus 4.7 lists the larger context window.

Benchmark Deep Dive

Agentic
BenchmarkClaude Opus 4.7Kimi K2.5Result
τ²-bench resultsSource 74%95.9%Kimi K2.5 leads
Gert LabsSource 65.59%45.88%Claude Opus 4.7 leads
ResearchClawBenchSource 20.7%14.0%Claude Opus 4.7 leads
OSWorld 2.0Source 13.9%—Not comparable
Terminal-Bench 2.0Source —50.8%Not comparable
BrowseCompSource —60.6%Not comparable
Claw-EvalSource —52.3%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
JobBenchSource —8.7%Not comparable
AA Agentic IndexSource —21.7%Not comparable
GDPval-AASource —25.4%Not comparable
GDPval-AASource —1009Not comparable
Coding
BenchmarkClaude Opus 4.7Kimi K2.5Result
Vibe Code BenchSource 71.00%—Not comparable
React Native EvalsSource 82.8%77.2%Claude Opus 4.7 leads
Terminal-Bench HardSource 54.5%34.8%Claude Opus 4.7 leads
AA-SciCodeSource 50.1%49.0%Claude Opus 4.7 leads
FrontierCodeSource 38.5%—Not comparable
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
SWE-RebenchSource —58.5%Not comparable
SciCodeSource —48.7%Not comparable
AA Coding IndexSource —46.8%Not comparable
Reasoning
BenchmarkClaude Opus 4.7Kimi K2.5Result
AA-LCRSource 67.0%65.3%Claude Opus 4.7 leads
CritPtSource 5.1%3.1%Claude Opus 4.7 leads
LongBench v2Source —61%Not comparable
Knowledge
BenchmarkClaude Opus 4.7Kimi K2.5Result
Artificial Analysis Intelligence IndexSource 42.7%35.4%Claude Opus 4.7 leads
AA-GPQA DiamondSource 88.5%87.9%Claude Opus 4.7 leads
AA-HLESource 31.2%29.4%Claude Opus 4.7 leads
AA-Omniscience IndexSource 14.2%-8.1%Claude Opus 4.7 leads
AA-Omniscience AccuracySource 43.5%34.3%Claude Opus 4.7 leads
AA-Omniscience Hallucination RateSource 51.9%64.6%Claude Opus 4.7 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
BenchmarkClaude Opus 4.7Kimi K2.5Result
FrontierMath v2 (Tiers 1-3)Source 43.793%27.900%Claude Opus 4.7 leads
FrontierMath v2 (Tier 4)Source 22.917%4.200%Claude Opus 4.7 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 Opus 4.7Kimi K2.5Result
MMLU-ProXSource —82.3%Not comparable
NOVA-63Source —56.0%Not comparable
Multimodal
BenchmarkClaude Opus 4.7Kimi K2.5Result
AA-MMMU-ProSource 76.4%75.4%Claude Opus 4.7 leads
Design Arena WebsiteSource 13281284Claude Opus 4.7 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 Opus 4.7Kimi K2.5Result
AA-IFBenchSource 43.6%70.2%Kimi K2.5 leads
IFEvalSource —93.9%Not comparable
Frequently Asked Questions (2)

Which is better, Claude Opus 4.7 or Kimi K2.5?

Claude Opus 4.7 is ahead on BenchLM's provisional leaderboard, 69 to 61. The biggest single separator in this matchup is FrontierMath v2 (Tier 4), where the scores are 22.917% and 4.200%.

Which is better for math, Claude Opus 4.7 or Kimi K2.5?

Kimi K2.5 has the edge for math in this comparison, averaging 60.6 versus 38.6. Inside this category, FrontierMath v2 (Tier 4) 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 Opus 4.7
API / mo$22,500
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 15, 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.