NOUMENOpen app
NOUMEN

Moonshot · Open weights

Kimi K2.6 Thinking

Added Apr 16, 2026Open weights

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Use in the app

Kimi K2.6 Thinking is the reasoning-optimized K2.6 variant for deeper multi-step planning and execution. It is tuned for long-horizon coding and design workflows, including complex orchestration across many specialized sub-agents and autonomous end-to-end output generation.

  • Reasoning
  • Tools
  • Vision
  • Parallel tools
Input
Text, Images
Output
Text
Discovery tier
Open weights

Pricing

USD per 1M tokens
Input
$0.45
Output
$1.89
Cache read
$0.0757

Reference prices as published by the model provider. API example

Specifications

Context window
262.1K
Max output
65.5K
Average output (7 days)
4K tokens
Longer than 98% of 465 models

Side by side

Kimi K2.6 Thinking compared

Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Kimi K2.6 ThinkingMoonshotThis modelKimi K3 TEEMoonshotRemove
ProviderMoonshotMoonshot
Discovery tierOpen weightsOpen weights
AddedApr 16, 2026Jul 31, 2026
Limits
Context window262.14K1.05MBest
Max output65.54KBest65.54K
Average output4K tokens1.1K tokens
Pricing, USD per 1M tokens
Input$0.45Best$3.00
Output$1.89Best$15.00
Cache read$0.0757Best$1.50
Performance
Throughput40.8 tok/sBest31.7 tok/s
Time to first token2.9 sBest5.0 s
Request success (24h)99.8%Best98.0%
Benchmark indexes
Intelligence Index27.043.6Best
Coding Index61.876.2Best
Capabilities
ReasoningYesYes
ToolsYesYes
VisionYesYes
Structured outputNoYes
Parallel toolsYesNo
Reasoning effort—low, high, max

Public evaluations

Benchmarks

Independent evaluation results, with how each score ranks among the models in the Noumen catalog.

  • Kimi K2.6 Thinking
  • Kimi K3 TEE

Intelligence Index

  • Kimi K2.6 Thinking27.0
  • Kimi K3 TEE43.6

Kimi K2.6 Thinking: Better than 68% of 326 scored models

3.8Catalog models by score55.1

Indexes

  • Coding Index

    Composite of the coding evaluations

    • Kimi K2.6 Thinking: 61.8
    • Kimi K3 TEE: 76.2Top

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • Kimi K2.6 Thinking: 13.0%
    • Kimi K3 TEE: 58.3%Top

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • Kimi K2.6 Thinking: 1.8%
    • Kimi K3 TEE: 24.8%Top

  • Harvey LAB-AAv1.1

    Legal tasks passed in full, with no material hallucination

    • Kimi K2.6 Thinking: —
    • Kimi K3 TEE: 5.3%

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • Kimi K2.6 Thinking: 819 Elo
    • Kimi K3 TEE: 1,499 EloTop

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • Kimi K2.6 Thinking: 1,038 Elo
    • Kimi K3 TEE: 1,534 EloTop

Document reasoning

  • GDP.pdf

    Professional reasoning over PDF documents, all-pass rate

    • Kimi K2.6 Thinking: 13.0%
    • Kimi K3 TEE: 22.0%Top

  • AA-LCRv1.1

    Reasoning across long documents

    • Kimi K2.6 Thinking: 81.0%
    • Kimi K3 TEE: 88.7%Top

  • MLCR-AA

    Long-context reasoning over medical records

    • Kimi K2.6 Thinking: 11.1%
    • Kimi K3 TEE: 21.7%Top

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • Kimi K2.6 Thinking: 37.5%
    • Kimi K3 TEE: 46.9%Top

  • GPQA Diamond

    Graduate-level science questions

    • Kimi K2.6 Thinking: 91.1%
    • Kimi K3 TEE: 93.5%Top

  • IFBench

    Following precise output instructions

    • Kimi K2.6 Thinking: 76.0%
    • Kimi K3 TEE: —

Coding

  • Terminal-Benchv4.0

    Practical software engineering in a terminal

    • Kimi K2.6 Thinking: 0.5%
    • Kimi K3 TEE: 12.6%Top

  • SciCode

    Scientific computing in Python

    • Kimi K2.6 Thinking: 51.5%
    • Kimi K3 TEE: 59.5%Top

Legacy evaluations

  • Terminal-Bench Hard

    Earlier terminal task suite

    • Kimi K2.6 Thinking: 43.9%
    • Kimi K3 TEE: —

  • τ²-Bench Telecom

    Tool use in a customer-support simulation

    • Kimi K2.6 Thinking: 95.9%
    • Kimi K3 TEE: —

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • Kimi K2.6 Thinking: 81.0%
    • Kimi K3 TEE: 88.7%Top

Scores from Artificial Analysis, updated Oct 26, 2026. Artificial Analysis

Live routing data

Performance and reliability

Recent averages across requests routed for this model. Your own speed depends on prompt length, reasoning effort and load.

Throughput
40.8 tok/s
Faster than 30% of 250 models
Time to first token
2.9 s
Quicker than 58% of 250 models
Prompt cache hits
51.7%
About $0.26 per 1M input tokens with caching
Request success (24h)
99.8%
Reliable

Cache hits are the share of input tokens recently served from a prompt cache across all requests. They are not a forecast of your own hit rate.

Hourly request success · last 72 hoursLowest hour 99.4%
  • Healthy (98% or more)
  • Degraded (90–98%)
  • Down (under 90%)
  • No requests

Developers

Use it through the API

Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.

Read the quickstart
curl https://www.noumen.xyz/api/v1/chat/completions \
  -H "Authorization: Bearer $NOUMEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "moonshotai/kimi-k2.6:thinking",
    "messages": [{ "role": "user", "content": "Explain this stack trace." }]
  }'
import os
from openai import OpenAI

client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
    model="moonshotai/kimi-k2.6:thinking",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)

Moonshot

Other Moonshot models, starting with the same family.