models/moonshotai/kimi-k2-6
M
MoonshotAI·active·free tier available

MoonshotAI: Kimi K2.6

MoonshotAI's flagship model. Context window: 262K tokens.

Overall score
4.62
/5.00 · ranked #22
Input
$0.600
per 1M tokens
Output
$3.41
per 1M tokens
Context
262K
tokens
Blended
$2.71
3:1 out:in ratio

Price drops, new benchmarks, model updates. Stay current on MoonshotAI: Kimi K2.6.

One email per change. Unsubscribe anytime.

modelpicker.aipowered by live benchmark data

Scores by test

Methodology →
Structured Output
5.0
Strategic Analysis
5.0
Constrained Rewriting
4.0
Creative Problem Solving
5.0
Tool Calling
4.0
Faithfulness
5.0
Classification
2.0
Long Context
5.0
Safety Calibration
5.0
Persona Consistency
5.0
Agentic Planning
5.0
Multilingual
5.0
Tabular Data
5.0
SWE-bench Verified
76.7
AIME 2025
96.1
GPQA Diamond
90.8
SciCode
53.5
APEX Agents
18.9
Epoch Capabilities Index (ECI)
151.0

What you need to know

Kimi K2.6 is a high-performance model optimized for complex reasoning and long-form data processing. Its primary differentiator is a near-perfect internal score across most technical domains, particularly in agentic planning, faithfulness, and structured output. With a 262K context window and a 5/5 score in long context handling, it is built for deep analysis of extensive datasets without losing coherence.

The model's pricing is competitive for its rank, with a blended cost of $2.81/MTok. Given its #5 overall ranking out of 71 models, the cost-to-performance ratio is high, providing frontier-level capabilities in strategic analysis and multilingual tasks at a mid-tier price point.

Despite its overall strength, the model has a significant deficiency in classification, scoring only 2/5. It also shows slight regressions in tool calling and constrained rewriting compared to its other capabilities. This indicates a model that excels at generative reasoning and synthesis but struggles with rigid categorization tasks.

Use this model for agentic workflows, complex strategic planning, and processing large documents where faithfulness is critical. Skip this model if your primary use case is high-accuracy text classification or strict categorical labeling.

Strengths — Top 3

Structured Output5.0/5.0
Strategic Analysis5.0/5.0
Creative Problem Solving5.0/5.0

Relative weaknesses — Bottom 3

Classification2.0/5.0
Constrained Rewriting4.0/5.0
Tool Calling4.0/5.0

Similar models

XMiMo-V2.5$0.2454.69MMeta: Muse Spark 1.1$3.504.77OOpenAI: GPT-5.6 Sol$23.754.77AAnthropic: Claude Opus 4.8 (Fast)$40.004.77