Skip to content
VECTOR WIREAI INTELLIGENCE
⌘K
UTC
Models741Deals30d261Policy30d65, 14 pendingSources1,013

Kimi K2.6

MoonshotReleased Apr 20, 2026Current

Score basis

Kimi K2.6 is strong in long context; capable in coding, multimodal tasks, instruction following, and reasoning; and behind the leaders in factuality, agentic tasks, and math. Too few results yet to rate safety or multilingual tasks.

Capability profile

Bars show the model's median result as a share of the leading model's, per capability. Select a row to see its results.

Strong1 capability
  1. Long Context−9.6%1 of 3
Capable4 capabilities
  1. Coding−18.2%10 of 10
  2. Multimodal−19.1%2 of 6
  3. Instruction Following−21.8%3 of 3
  4. Reasoning−24.1%4 of 6
Limited3 capabilities
  1. Factuality−26.4%4 of 4
  2. Agentic−29.5%6 of 7
  3. Math−42.1%3 of 5
Not rated2 capabilities

Too few results yet to rate Safety or Multilingual.

Price

$0.95input$4.00outputper million tokens

From Artificial Analysis · 5 providers tracked · All prices

Costs more than 68% of 331 priced models · 3:1 input-to-output blend, log scale

Evidence

181results on116benchmarks

  • 26 independently verified
  • 36 aggregator
  • 44 vendor-reported
  • 75 cross-referenced

From 24 sources · latest Oct 8, 2026 · How verification works

API features

Tool callingStructured outputsReasoning

As listed by OpenRouter

Research

17 papers reference Kimi K2.6

Kimi K2.6 benchmark results

181 results on 116 benchmarks, grouped by capability. Each bar is the result as a share of the capability leader's; every result links to its source.

Long Context

StrongFull long context ranking

9.6% behind the leader1 of 3 ranked benchmarks measured

Show 1 more long context resultHide 1 long context result

Coding

CapableFull coding ranking

18.2% behind the leader10 of 10 ranked benchmarks measured

Show 12 more coding resultsHide 12 coding results

Multimodal

CapableFull multimodal ranking

19.1% behind the leader2 of 6 ranked benchmarks measured

Show 6 more multimodal resultsHide 6 multimodal results

Instruction Following

CapableFull instruction following ranking

21.8% behind the leader3 of 3 ranked benchmarks measured

Show 2 more instruction following resultsHide 2 instruction following results

Reasoning

CapableFull reasoning ranking

24.1% behind the leader4 of 6 ranked benchmarks measured

Show 13 more reasoning resultsHide 13 reasoning results

Factuality

LimitedFull factuality ranking

26.4% behind the leader4 of 4 ranked benchmarks measured

Show 4 more factuality resultsHide 4 factuality results

Agentic

LimitedFull agentic ranking

29.5% behind the leader6 of 7 ranked benchmarks measured

Show 9 more agentic resultsHide 9 agentic results

Math

LimitedFull math ranking

42.1% behind the leader3 of 5 ranked benchmarks measured

Show 11 more math resultsHide 11 math results

More results

Benchmarks outside the capability baskets. They are not ranked against a leader.

Show 84 more resultsHide 84 results

Kimi K2.6: common questions

Who makes Kimi K2.6?

Kimi K2.6 is made by Moonshot.

When was Kimi K2.6 released?

Kimi K2.6 was released on Apr 20, 2026, according to Artificial Analysis.

What is Kimi K2.6 good at?

Kimi K2.6 is strong in long context; capable in coding, multimodal tasks, instruction following, and reasoning; and behind the leaders in factuality, agentic tasks, and math. Too few results yet to rate safety or multilingual tasks.

How much does Kimi K2.6 cost?

Kimi K2.6 costs $0.95 per million input tokens and $4.00 per million output tokens, according to Artificial Analysis. We track its price at 5 providers. At a mix of three input tokens to one output token, it costs more than 68% of the 331 priced models we track.

How many benchmarks has Kimi K2.6 been tested on?

We track 181 results for Kimi K2.6 on 116 benchmarks from 24 sources, 26 of them independently verified. The latest was recorded on Oct 8, 2026.

Which API features does Kimi K2.6 support?

OpenRouter lists tool calling, structured outputs, and reasoning for Kimi K2.6.

About this record

Where Kimi K2.6's numbers come from, and every name it appears under.

Tracked since
May 2, 2026
Newest source mention
Aug 23, 2026

Where the results come from

Verification: 181 scores · 26 independently verified · 36 aggregator-attributed · 75 vendor cross-reference · 44 vendor-reported. How these tiers are assigned

From 24 sources on 11 sites. Hugging Face supplies 89 of them; the 26 independently verified results come from 8 sites. Bars are coloured by trust tier.

  • huggingface.co89
  • artificialanalysis.ai37
  • api.llm-stats.com26
  • livebench.ai7
  • matharena.ai5
  • epoch.ai4
  • raw.githubusercontent.com4
  • thinkingmachines.ai4
  • datasets-server.huggingface.co2
  • lmarena.ai2
  • labs.scale.com1

Also known as

Kimi K2.6 (Think)K2.6 ThinkingKimi K2.6 (Non-Reasoning)kimi-k2-6kimi-k2.6 (w/o explore)kimi-k2.6 w/o explorekimi k2.6 (reasoning)kimi-k2-6-non-reasoningkimi k 2.6moonshot k2.6moonshotai/Kimi-K2.6Kimi K2.6 1T A32B