Kimi K2.5 Benchmark Details

Kimi K2.5 currently shows benchmark results led by LiveCodeBench (19 / 128, score 85), HLE (36 / 197, score 50.20), AIME2025 (21 / 107, score 96.10). This page also compares it with 2 competitor models and 3 predecessor or same-series models, including performance and pricing views when available. 1 source link is attached for reference.

Benchmark Results

Kimi K2.5

Benchmark Results

General Knowledge

6 evaluations

Benchmark / mode

Score

Rank/total

MMLU Pro

Thinking Mode

78.50

70 / 134

LiveBench

Thinking Mode

69.07

41 / 117

ARC-AGI-1

Thinking Mode

65.30

58 / 92

HLE

Thinking Mode

30.10

109 / 197

HLE

Thinking ModeTools

50.20

36 / 197

ARC-AGI-2

Thinking Mode

11.80

62 / 85

Other

1 evaluations

Benchmark / mode

Score

Rank/total

GPQA Diamond

Thinking Mode

87.60

78 / 274

Coding and Software Engineer

6 evaluations

Benchmark / mode

Score

Rank/total

Text Arena (Coding)

Standard Mode

1430.56

27 / 35

LiveCodeBench

Thinking Mode

19 / 128

SWE-bench Verified

Thinking ModeTools

76.80

30 / 116

SWE-bench Multilingual

Thinking Mode

17 / 29

SWE-Bench Pro - Public

Thinking ModeTools

50.70

47 / 62

WeirdML v2

Standard ModeTools

45.60

42 / 52

Math and Reasoning

4 evaluations

Benchmark / mode

Score

Rank/total

AIME2025

Thinking Mode

96.10

21 / 107

AIME 2026

Thinking Mode

92.50

13 / 21

IMO-AnswerBench

Thinking Mode

81.80

18 / 24

FrontierMath - Tier 4

Standard Mode

4.20

40 / 80

Writing and Creative Capabilities

1 evaluations

Benchmark / mode

Score

Rank/total

Creative Writing

Standard Mode

1575.80

43 / 106

Common Sense Reasoning

1 evaluations

Benchmark / mode

Score

Rank/total

SimpleBench

Thinking Mode

46.80

54 / 94

AI Agent - Information Search

1 evaluations

Benchmark / mode

Score

Rank/total

BrowseComp

Thinking ModeToolsInternet

60.60

38 / 57

AI Agent - Tool Usage

2 evaluations

Benchmark / mode

Score

Rank/total

MCP-Atlas

Standard ModeTools

64.40

32 / 41

Terminal Bench 2.0

Thinking ModeTools

50.80

35 / 48

Text Embedding

2 evaluations

Benchmark / mode

Score

Rank/total

Context Arena

Standard Mode

53.33

83 / 126

Context Arena

Thinking Mode

59.22

77 / 126

Productivity Knowledge

1 evaluations

Benchmark / mode

Score

Rank/total

GDPval-AA

Thinking Mode

15 / 21

Long Context

2 evaluations

Benchmark / mode

Score

Rank/total

AA-LCR

Thinking Mode

22 / 29

LongBench v2

Standard Mode

6 / 14

Claw-style Agent Evaluation

2 evaluations

Benchmark / mode

Score

Rank/total

Pinch Bench

Thinking ModeTools

84.80

18 / 38

Claw Bench

Thinking ModeTools

81.70

18 / 29

Other

1 evaluations

Benchmark / mode

Score

Rank/total

Fiction.liveBench

Standard Mode

86.10

7 / 16

Compare with other models

Competitor Comparison

Benchmark scores for Kimi K2.5 compared against top models in its class

Kimi K2.5GLM-5 MiniMax M2.5

Benchmark categories:

The chart shows each model’s highest score per benchmark within the current filter. Out-of-100 benchmarks use raw heights; out-of-range benchmarks are scaled within that benchmark while labels keep the original scores.

12 benchmarks with comparable scores. Each model shows its best score; mode label is displayed below.

Benchmark	Kimi K2.5Current	GLM-5	MiniMax M2.5
ARC-AGI-1 Score (%) 综合评估	65.30Thinking Enabled	44.70Thinking Enabled	63.70Thinking Enabled
ARC-AGI-2 Accuracy and cost per task 综合评估	11.80Thinking Enabled	4.90Thinking Enabled	4.90Thinking Enabled
HLE Accuracy 综合评估	50.20Thinking Enabled ｜ Tools	50.40Thinking Enabled ｜ Tools	19.40Thinking Enabled
LiveBench Accuracy 综合评估	69.07Thinking Enabled	68.85Standard Mode	60.14Deep Thinking Mode
GPQA Diamond Accuracy 科学与综合推理	87.60Thinking Enabled	86.00Thinking Enabled	85.20Thinking Enabled
SWE-Bench Pro - Public Accuracy 编程与软件工程	50.70Thinking Enabled ｜ Tools	--	55.40Thinking Enabled ｜ Tools
SWE-bench Verified Accuracy 编程与软件工程	76.80Thinking Enabled ｜ Tools	77.80Thinking Enabled	80.20Thinking Enabled ｜ Tools
AIME 2026 Accuracy 数学推理	92.50Thinking Enabled	92.70Thinking Enabled	--
AIME2025 Accuracy 数学推理	96.10Thinking Enabled	--	86.30Thinking Enabled
FrontierMath - Tier 4 Accuracy 数学推理	4.20Standard Mode	2.10Standard Mode	--
IMO-AnswerBench Accuracy 数学推理	81.80Thinking Enabled	82.50Thinking Enabled	--
Creative Writing Elo、大模型评判两两对战写作和创作	1575.80Standard Mode	1597.50Standard Mode	1358.40Standard Mode

8 additional benchmarks remain in the chart above.

Standard API Pricing: Kimi K2.5 vs. Peer Models

Shows standard text input and output pricing side by side for each model. If extended-context pricing exists, the chart keeps the base rate and explains the threshold below.

Source: DataLearnerAI. Standard text prices shown here use the default supplier. · USD / 1M tokens

Model	Supplier	Standard input	Standard output	Base price applies to
Kimi K2.5	Moonshot AI	$0.6 / 1M tokens	$3 / 1M tokens	—
GLM-5	智谱AI	$1 / 1M tokens	$3.2 / 1M tokens	—
MiniMax M2.5	MiniMaxAI	$0.3 / 1M tokens	$2.4 / 1M tokens	—

Version History

How each version of the Kimi K2.5 series stacks up on benchmark tests

Kimi K2.5Kimi K2 Thinking Kimi K2 0905 Kimi K2

Benchmark categories:

12 benchmarks with comparable scores. Each model shows its best score; mode label is displayed below.· Click a row to view its trend chart.

Benchmark	Kimi K2.5Current	Kimi K2 Thinking	Kimi K2 0905	Kimi K2
ARC-AGI-1 Score (%) 综合评估	65.30Thinking Enabled	--	--	13.30Standard Mode
HLE Accuracy 综合评估	50.20Thinking Enabled ｜ Tools	51.00Thinking Enabled ｜ Tools	21.70Thinking Enabled ｜ Tools	4.70Standard Mode
LiveBench Accuracy 综合评估	69.07Thinking Enabled	61.59Thinking Enabled	--	48.10Standard Mode
MMLU Pro Accuracy 综合评估	78.50Thinking Enabled	84.60Thinking Enabled	--	81.10Standard Mode
GPQA Diamond Accuracy 科学与综合推理	87.60Thinking Enabled	84.50Thinking Enabled	--	75.10Standard Mode
LiveCodeBench Pass @K 编程与软件工程	85.00Thinking Enabled	83.10Thinking Enabled	--	53.70Standard Mode
SWE-Bench Pro - Public Accuracy 编程与软件工程	50.70Thinking Enabled ｜ Tools	--	27.67Standard Mode	--
SWE-bench Verified Accuracy 编程与软件工程	76.80Thinking Enabled ｜ Tools	71.30Thinking Enabled ｜ Tools	69.20Standard Mode	51.80Standard Mode
AIME2025 Accuracy 数学推理	96.10Thinking Enabled	100.00Thinking Enabled ｜ Tools	75.20Thinking Enabled ｜ Tools	54.00Standard Mode
FrontierMath - Tier 4 Accuracy 数学推理	4.20Standard Mode	0.00Thinking Enabled	--	0.01Standard Mode
Creative Writing Elo、大模型评判两两对战写作和创作	1575.80Standard Mode	1627.90Standard Mode	--	1662.70Standard Mode
SimpleBench Score (AVG@5) 常识推理	46.80Thinking Enabled	39.60Standard Mode	--	26.30Standard Mode

2 additional benchmarks remain in the chart above.

Single-Benchmark Version Trend

Viewing: ARC-AGI-1 · 综合评估

Benchmark

NormalNormal + ToolsThinkingThinking + ToolsDeepDeep + Tools

X-axis shows model and release date, Y-axis shows score; solid lines connect the same mode across versions, while dotted guides align modes within the same generation.

Standard API Pricing Across the Kimi K2.5 Series

Shows standard text input and output pricing side by side for each model. If extended-context pricing exists, the chart keeps the base rate and explains the threshold below.

Source: DataLearnerAI. Standard text prices shown here use the default supplier. · USD / 1M tokens

Model	Supplier	Standard input	Standard output	Base price applies to
Kimi K2.5	Moonshot AI	$0.6 / 1M tokens	$3 / 1M tokens	—
Kimi K2 Thinking	Fireworks AI	$0.6 / 1M tokens	$2.5 / 1M tokens	—
Kimi K2 0905	Fireworks AI	$0.6 / 1M tokens	$2.5 / 1M tokens	—
Kimi K2	Moonshot AI	$0.6 / 1M tokens	$2.5 / 1M tokens	—

Sources

kimi.comkimi.com