Skip to main content
BenchLM
Advertise

Model comparison

GLM-5 vs Kimi K2.5

Data verified

Head-to-head evidence from 47 shared benchmark results across 8 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

GLM-5
Z.AI
66.06/100
Margin
6.4pts
← winning
Kimi K2.5
Moonshot AI
59.66/100
4 category wins3 category wins

Public leaderboard positions: GLM-5 #28 (Supported); Kimi K2.5 #54 (Supported). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.

Evidence parity. GLM-5 and Kimi K2.5 share 47 comparable benchmark results. 7 of 8 categories are comparable. 2 results are unique to GLM-5; 16 to Kimi K2.5.

Updated July 20, 2026
Shared results
47
GLM-5 only
2
Kimi K2.5 only
16
Comparable categories
7 / 8

Pick GLM-5 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 47 shared benchmark results across 8 evidence categories; 7 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.

Why this result

GLM-5 is clearly ahead on the BenchAlign aggregate, 66.06 to 59.66. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.

GLM-5's sharpest advantage is in knowledge, where it averages 66.4 against 56.9. The single biggest benchmark swing on the page is HLE, 50.4% to 30.1%. Kimi K2.5 does hit back in mathematics, so the answer changes if that is the part of the workload you care about most.

GLM-5 is also the more expensive model on tokens at $1.00 input / $3.20 output per 1M tokens, versus $0.60 input / $3.00 output per 1M tokens for Kimi K2.5. Kimi K2.5 gives you the larger context window at 256K, compared with 200K for GLM-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 GLM-5 and Kimi K2.5
CategoryGLM-5ΔKimi K2.5
KnowledgeGLM-566.4Margin← 9.5Kimi K2.556.9
CodingGLM-566.3Margin← 6.9Kimi K2.559.4
MathGLM-556.3Margin→ 4.3Kimi K2.560.6
Inst. FollowingGLM-592.6Margin→ 1.3Kimi K2.593.9
AgenticGLM-556.2Margin← 1.2Kimi K2.555.0
MultilingualGLM-583.1Margin← 0.8Kimi K2.582.3
ReasoningGLM-560.8Margin→ 0.2Kimi K2.561.0
MultimodalGLM-5Not measuredMarginNo overlapKimi K2.578.5

Decisive benchmark drivers

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

More
A · GLM-5B · Kimi K2.5

  1. HLE

    Knowledge
    Source ↗
    A 50.4%B 30.1%
    Winner: GLM-5Δ 20.3
    HLE: GLM-5 scored 50.4%; Kimi K2.5 scored 30.1%. GLM-5 wins this benchmark.

  2. FrontierMath v2 (Tiers 1-3)

    Math
    Source ↗
    A 16.434%B 27.900%
    Winner: Kimi K2.5Δ 11.5
    FrontierMath v2 (Tiers 1-3): GLM-5 scored 16.434%; Kimi K2.5 scored 27.900%. Kimi K2.5 wins this benchmark.

  3. Terminal-Bench 2.0

    Agentic
    Source ↗
    A 56.2%B 50.8%
    Winner: GLM-5Δ 5.4
    Terminal-Bench 2.0: GLM-5 scored 56.2%; Kimi K2.5 scored 50.8%. GLM-5 wins this benchmark.

  4. SWE-bench Pro

    Coding
    Source ↗
    A 55.1%B 50.7%
    Winner: GLM-5Δ 4.4
    SWE-bench Pro: GLM-5 scored 55.1%; Kimi K2.5 scored 50.7%. GLM-5 wins this benchmark.

  5. SWE-Rebench

    Coding
    Source ↗
    A 62.8%B 58.5%
    Winner: GLM-5Δ 4.3
    SWE-Rebench: GLM-5 scored 62.8%; Kimi K2.5 scored 58.5%. GLM-5 wins this benchmark.

Operational comparison

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

MetricGLM-5Kimi K2.5Comparison
Input / output priceUSD per 1M tokensGLM-5$1 input / $3.2 outputKimi K2.5$0.6 input / $3 outputKimi K2.5 has the lower combined listed price.
Generation speedtokens per secondGLM-574 tok/sKimi K2.545 tok/sGLM-5 has the higher measured throughput.
First-answer latencyseconds to first tokenGLM-51.64 sKimi K2.52.38 sGLM-5 reaches the first token sooner.
Context windowmaximum listed tokensGLM-5200KKimi K2.5256KKimi K2.5 lists the larger context window.

Benchmark Deep Dive

AgenticGLM-5 wins
BenchmarkGLM-5Kimi K2.5Result
Terminal-Bench 2.0Source 56.2%50.8%GLM-5 leads
Claw-EvalSource 57.7%52.3%GLM-5 leads
QwenClawBenchSource 54.1%54.3%Kimi K2.5 leads
τ³-bench resultsSource 65.6%65.7%Kimi K2.5 leads
DeepPlanningSource 14.6%14.4%GLM-5 leads
ToolathlonSource 38%27.8%GLM-5 leads
MCP AtlasSource 31.1%29.5%GLM-5 leads
MCP-TasksSource 60.8%59.1%GLM-5 leads
WideResearchSource 69.8%72.7%Kimi K2.5 leads
τ²-bench resultsSource 98.2%95.9%GLM-5 leads
CyberGymSource 43.2%—Not comparable
APEX-Agents-AASource 14.5%11.5%GLM-5 leads
Gert LabsSource 50.99%45.88%GLM-5 leads
BrowseCompSource —60.6%Not comparable
DeepSearchQASource —77.1%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
CodingGLM-5 wins
BenchmarkGLM-5Kimi K2.5Result
SWE-bench VerifiedSource 77.8%76.8%GLM-5 leads
SWE-bench Verified*Source 72.8%70.8%GLM-5 leads
SWE-bench ProSource 55.1%50.7%GLM-5 leads
SWE MultilingualSource 73.3%73%GLM-5 leads
SWE-RebenchSource 62.8%58.5%GLM-5 leads
React Native EvalsSource 74.8%77.2%Kimi K2.5 leads
AA-SciCodeSource 46.2%49.0%Kimi K2.5 leads
LiveCodeBench v6Source —85.0%Not comparable
SciCodeSource —48.7%Not comparable
AA Coding IndexSource —46.8%Not comparable
ReasoningKimi K2.5 wins
BenchmarkGLM-5Kimi K2.5Result
LongBench v2Source 60.8%61%Kimi K2.5 leads
AI-NeedleSource 63.3%—Not comparable
AA-LCRSource 63.3%65.3%Kimi K2.5 leads
CritPtSource 2.0%3.1%Kimi K2.5 leads
KnowledgeGLM-5 wins
BenchmarkGLM-5Kimi K2.5Result
GPQASource 86%87.6%Kimi K2.5 leads
GPQA-DSource 86.0%87.6%Kimi K2.5 leads
SuperGPQASource 66.8%69.2%Kimi K2.5 leads
MMLU-ProSource 85.7%87.1%Kimi K2.5 leads
MMLU-Pro (Arcee)Source 85.8%87.1%Kimi K2.5 leads
HLESource 50.4%30.1%GLM-5 leads
Artificial Analysis Intelligence IndexSource 39.5%35.4%GLM-5 leads
AA-GPQA DiamondSource 82.0%87.9%Kimi K2.5 leads
AA-HLESource 27.2%29.4%Kimi K2.5 leads
AA-Omniscience IndexSource 2.0%-8.1%GLM-5 leads
AA-Omniscience AccuracySource 26.9%34.3%Kimi K2.5 leads
AA-Omniscience Hallucination RateSource 34.0%64.6%GLM-5 leads
MathKimi K2.5 wins
BenchmarkGLM-5Kimi K2.5Result
AIME26Source 95.8%95.8%Tie
AIME25 (Arcee)Source 93.3%96.3%Kimi K2.5 leads
HMMT Feb 2025Source 97.5%95.4%GLM-5 leads
HMMT Nov 2025Source 96.9%91.1%GLM-5 leads
HMMT Feb 2026Source 86.4%87.1%Kimi K2.5 leads
MMAnswerBenchSource 82.5%81.8%GLM-5 leads
FrontierMath v2 (Tiers 1-3)Source 16.434%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
MultilingualGLM-5 wins
BenchmarkGLM-5Kimi K2.5Result
MMLU-ProXSource 83.1%82.3%GLM-5 leads
NOVA-63Source 55.1%56.0%Kimi K2.5 leads
Multimodal
BenchmarkGLM-5Kimi K2.5Result
Design Arena WebsiteSource 12801282Kimi 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. FollowingKimi K2.5 wins
BenchmarkGLM-5Kimi K2.5Result
IFEvalSource 92.6%93.9%Kimi K2.5 leads
AA-IFBenchSource 72.3%70.2%GLM-5 leads
Frequently Asked Questions (8)

Which is better, GLM-5 or Kimi K2.5?

GLM-5 is ahead on BenchLM's BenchAlign leaderboard, 66.06 to 59.66. The biggest single separator in this matchup is HLE, where the scores are 50.4% and 30.1%.

Which is better for knowledge tasks, GLM-5 or Kimi K2.5?

GLM-5 has the edge for knowledge tasks in this comparison, averaging 66.4 versus 56.9. Inside this category, AA-Omniscience Hallucination Rate is the benchmark that creates the most daylight between them.

Which is better for coding, GLM-5 or Kimi K2.5?

GLM-5 has the edge for coding in this comparison, averaging 66.3 versus 59.4. Inside this category, SWE-bench Pro is the benchmark that creates the most daylight between them.

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

Kimi K2.5 has the edge for math in this comparison, averaging 60.6 versus 56.3. Inside this category, FrontierMath v2 (Tiers 1-3) is the benchmark that creates the most daylight between them.

Which is better for reasoning, GLM-5 or Kimi K2.5?

Kimi K2.5 has the edge for reasoning in this comparison, averaging 61 versus 60.8. Inside this category, AA-LCR is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, GLM-5 or Kimi K2.5?

GLM-5 has the edge for agentic tasks in this comparison, averaging 56.2 versus 55. Inside this category, Toolathlon is the benchmark that creates the most daylight between them.

Which is better for instruction following, GLM-5 or Kimi K2.5?

Kimi K2.5 has the edge for instruction following in this comparison, averaging 93.9 versus 92.6. Inside this category, AA-IFBench is the benchmark that creates the most daylight between them.

Which is better for multilingual tasks, GLM-5 or Kimi K2.5?

GLM-5 has the edge for multilingual tasks in this comparison, averaging 83.1 versus 82.3. Inside this category, NOVA-63 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.

GLM-5
API / mo$3,150
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

Z.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.