Kimi K2 Thinking

Moonshot AI's distinct Kimi K2 Thinking checkpoint is a mandatory-reasoning model for long-horizon agentic workflows and multi-step tool use.

Reasoning
Tool Calling
Structured Output

Added Nov 6, 2025

Model weights

Pricing

Auto routing · per 1M tokens

Input: $0.60
Output: $2.50
Cache read: $0.15

Compare provider prices

Specifications

Context window: 262.1K
Max output: 98.3K
Parameters: 1T / 32B; Total / active
Avg output (7d): 1.8K tokens; Longer than 88% of models

Benchmarks

Artificial Analysis

Design Arena

Sourced from Artificial Analysis.

Intelligence Index

22.0

Better than 68% of models compared

Agentic work

T²-Bench Telecom (legacy)

Legacy fallback · Conversational AI agents in dual-control scenarios

93.0%

Better than 89% of models compared

Document reasoning

AA-LCR v1.1

Long context reasoning with updated grading

72.0%

Better than 67% of models compared

Reasoning

HLE

Humanity's Last Exam

23.8%

Better than 72% of models compared

IFBench

Instruction-following benchmark

68.1%

Better than 81% of models compared

Coding

Terminal-Bench Hard (legacy)

Legacy fallback · Agentic coding and terminal use

31.1%

Better than 73% of models compared

LiveCodeBench

Contamination-free coding benchmark

85.3%

Better than 96% of models compared

Math

AIME 2025

American Invitational Mathematics Examination 2025

94.7%

Better than 96% of models compared

Knowledge

MMLU-Pro

Professional and academic subject knowledge

84.8%

Better than 89% of models compared

Legacy benchmarks

GPQA Diamond (legacy)

Graduate-level scientific reasoning

83.8%

Better than 74% of models compared

AA-LCR (unversioned / legacy)

Long context reasoning evaluation