Skip to main content
BenchLM
Advertise

Model comparison

Kimi K2.5 vs o3

Data verified

Head-to-head evidence from 15 shared benchmark results across 7 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

Kimi K2.5
Moonshot AI
59.66/100
Margin
11.8pts
← winning
o3
OpenAI
47.89/100
1 category wins0 category wins

Public leaderboard positions: Kimi K2.5 #54 (Supported); o3 #131 (Supported). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.

Evidence parity. Kimi K2.5 and o3 share 15 comparable benchmark results. 1 of 8 categories are comparable. 48 results are unique to Kimi K2.5; 1 to o3.

Updated July 20, 2026
Shared results
15
Kimi K2.5 only
48
o3 only
1
Comparable categories
1 / 8

Pick Kimi K2.5 if you want the stronger benchmark profile. o3 only becomes the better choice if you want the stronger reasoning-first profile.

Confidence note. This is a partial-evidence comparison with 15 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

Kimi K2.5 is clearly ahead on the BenchAlign aggregate, 59.66 to 47.89. 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 14.5. The single biggest benchmark swing on the page is FrontierMath v2 (Tiers 1-3), 27.900% to 18.685%.

o3 is also the more expensive model on tokens at $2.00 input / $8.00 output per 1M tokens, versus $0.60 input / $3.00 output per 1M tokens for Kimi K2.5. That is roughly 2.7x on output cost alone. o3 is the reasoning model in the pair, while Kimi K2.5 is not. That usually helps on harder chain-of-thought-heavy tests, but it can also mean more latency and more token spend in real use. Kimi K2.5 gives you the larger context window at 256K, compared with 200K for o3.

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 o3
CategoryKimi K2.5Δo3
MathKimi K2.560.6Margin← 46.1o314.5
AgenticKimi K2.555.0MarginNo overlapo3Not measured
CodingKimi K2.559.4MarginNo overlapo3Not measured
ReasoningKimi K2.561.0MarginNo overlapo3Not measured
KnowledgeKimi K2.556.9MarginNo overlapo3Not measured
MultilingualKimi K2.582.3MarginNo overlapo3Not measured
MultimodalKimi K2.578.5MarginNo overlapo3Not measured
Inst. FollowingKimi K2.593.9MarginNo overlapo3Not measured

Decisive benchmark drivers

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

More
A · Kimi K2.5B · o3

  1. FrontierMath v2 (Tiers 1-3)

    Math
    Source ↗
    A 27.900%B 18.685%
    Winner: Kimi K2.5Δ 9.2
    FrontierMath v2 (Tiers 1-3): Kimi K2.5 scored 27.900%; o3 scored 18.685%. Kimi K2.5 wins this benchmark.

  2. FrontierMath v2 (Tier 4)

    Math
    Source ↗
    A 4.200%B 2.083%
    Winner: Kimi K2.5Δ 2.1
    FrontierMath v2 (Tier 4): Kimi K2.5 scored 4.200%; o3 scored 2.083%. Kimi K2.5 wins this benchmark.

Operational comparison

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

MetricKimi K2.5o3Comparison
Input / output priceUSD per 1M tokensKimi K2.5$0.6 input / $3 outputo3$2 input / $8 outputKimi K2.5 has the lower combined listed price.
Generation speedtokens per secondKimi K2.545 tok/so3118 tok/so3 has the higher measured throughput.
First-answer latencyseconds to first tokenKimi K2.52.38 so35.38 sKimi K2.5 reaches the first token sooner.
Context windowmaximum listed tokensKimi K2.5256Ko3200KKimi K2.5 lists the larger context window.

Benchmark Deep Dive

Agentic
BenchmarkKimi K2.5o3Result
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
τ²-bench resultsSource 95.9%80.7%Kimi K2.5 leads
APEX-Agents-AASource 11.5%—Not comparable
Gert LabsSource 45.88%—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 1009—Not comparable
Coding
BenchmarkKimi K2.5o3Result
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
React Native EvalsSource 77.2%—Not comparable
SciCodeSource 48.7%—Not comparable
AA-SciCodeSource 49.0%41.0%Kimi K2.5 leads
AA Coding IndexSource 46.8%—Not comparable
Reasoning
BenchmarkKimi K2.5o3Result
LongBench v2Source 61%—Not comparable
AA-LCRSource 65.3%69.3%o3 leads
CritPtSource 3.1%1.1%Kimi K2.5 leads
Knowledge
BenchmarkKimi K2.5o3Result
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%30.4%Kimi K2.5 leads
AA-GPQA DiamondSource 87.9%82.7%Kimi K2.5 leads
AA-HLESource 29.4%20.0%Kimi K2.5 leads
AA-Omniscience IndexSource -8.1%-15.3%Kimi K2.5 leads
AA-Omniscience AccuracySource 34.3%38.4%o3 leads
AA-Omniscience Hallucination RateSource 64.6%87.1%Kimi K2.5 leads
MathKimi K2.5 wins
BenchmarkKimi K2.5o3Result
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%18.685%Kimi K2.5 leads
FrontierMath v2 (Tier 4)Source 4.200%2.083%Kimi K2.5 leads
AA MATH-500Source —99.2%Not comparable
Multilingual
BenchmarkKimi K2.5o3Result
MMLU-ProXSource 82.3%—Not comparable
NOVA-63Source 56.0%—Not comparable
Multimodal
BenchmarkKimi K2.5o3Result
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%70.1%Kimi K2.5 leads
Design Arena WebsiteSource 12821069Kimi K2.5 leads
Inst. Following
BenchmarkKimi K2.5o3Result
IFEvalSource 93.9%—Not comparable
AA-IFBenchSource 70.2%71.4%o3 leads
Frequently Asked Questions (2)

Which is better, Kimi K2.5 or o3?

Kimi K2.5 is ahead on BenchLM's BenchAlign leaderboard, 59.66 to 47.89. The biggest single separator in this matchup is FrontierMath v2 (Tiers 1-3), where the scores are 27.900% and 18.685%.

Which is better for math, Kimi K2.5 or o3?

Kimi K2.5 has the edge for math in this comparison, averaging 60.6 versus 14.5. 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.

Kimi K2.5
API / mo$2,700
Self-host / mo$5,221
Break-even132M/day
o3
API / mo$7,500
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 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.