Moonshot AI
Basic

Kimi K2 Thinking

kimi-k2-thinking

Moonshot AI's thinking mode — deep reflection before answering for complex reasoning on Basic.

ReasoningThinking

Technical Specifications

API Identifierkimi-k2-thinking
PlansBasic
Maximum ContextNon communiqué
Vision (images)No
Native Web AccessNo
Image GenerationNo
Input Cost$0.55 / M tokens
Output Cost$2.50 / M tokens

Overview

Kimi K2 Thinking is the thinking-mode version of Kimi K2, available on Basic on Mingo. Thinking mode lets the model 'think out loud' before providing an answer, producing deeper and more transparent reasoning on complex problems. This is the key difference from K2 Instruct: slower, but more rigorous for demanding tasks.


Strengths

Thinking modeRigorous reasoningTransparent

Cas d'usage

Complex reasoning

Problems requiring in-depth, structured multi-step reflection.

Argument analysis

Critical evaluation of complex arguments with methodical reflection.

STEM problem solving

Science, mathematics, and technology with verifiable reasoning.

Pricing

Input Tokens
$0.55
per million tokens
Output Tokens
$2.50
per million tokens

Frequently Asked Questions

Is thinking mode slower than Instruct?
Yes, thinking mode takes longer because the model thinks before answering. This extra delay is offset by better reasoning quality on complex problems.
Is Kimi K2 Thinking comparable to Claude Sonnet for reasoning on Basic?
Kimi K2 Thinking is an interesting alternative for pure reasoning. Claude Sonnet 4.6 remains more versatile (vision, 1M tokens), but Kimi K2 Thinking can outperform it on specific reasoning tasks.

Think with Kimi K2 Thinking on Basic

Thinking mode available on Basic — deep and structured reasoning.

Try Mingo Basic