// COMPARISON

Kimi K2 Thinking

A head-to-head benchmark comparison of Kimi K2 Thinking across 27 evaluations.

The models

MO

Kimi K2 Thinking

Moonshot
Language Open (Restricted)
Params
1T
Context
β€”
Released
Nov 2025

Today, we are introducing Kimi K2 Thinking, our best open-source thinking model. Built as a thinking agent, it reasons step by step while using tools, achieving.

Full Kimi K2 Thinking specs β†’

Benchmark comparison

Benchmark Kimi K2 Thinking
agentic
BrowseComp 60.2%
BrowseComp-Zh 62.3
FinSearchComp-T3 47.4
MCP-Mark 20.4
MCP-Universe 35.6
Seal-0 56.3
Terminal-Bench 2.0 47.1%
Tool-Decathlon 17.6
τ²-Bench (TAU2) 74.3
code
LiveCodeBench 82.6%
LiveCodeBench v6 83.1%
SWE-bench Verified 71.3%
SciCode 44.8%
coding
Multi-SWE-bench 41.9
OJ-Bench (cpp) 48.7
SWE-Bench Multilingual 61.1
general
Longform Writing 73.8
MMLU-Pro 84.6%
MMLU-Redux 94.4%
long-context
Frames 87
math
AIME 2025 100%
HMMT February 2025 97.5%
HMMT November 2025 89.2
IMOAnswerBench 78.6
reasoning
GPQA Diamond 84.5%
Humanity's Last Exam 51%
safety
HealthBench 58%

Best result per row highlighted in cyan. Each benchmark links to its definition and sources; each model links to its full scorecard.

Frequently asked questions

Which of these models is best for coding? β–Ά

Kimi K2 Thinking has the strongest result on LiveCodeBench among the models compared here. See the code-category rows in the table for the full picture.

How many benchmarks are compared? β–Ά

This comparison covers 27 benchmarks on which at least one of the selected models has a published score.

Where do the benchmark scores come from? β–Ά

Scores are aggregated from official model cards, technical reports and standard public evaluations, and link back to each benchmark's source. They are updated as new results are published.

Explore further

β†’ Kimi K2 Thinking β†’ AI benchmarks leaderboard β†’ Build a different comparison
Build a different comparison

Models (0 selected)

Benchmarks (auto)

Select at least two models.