Skip to content
aiqbench

Moonshot · model passport

MoonshotAI: Kimi K2.6

kimi-k2-6
  • Open weights
  • Modified MIT licence
  • 1.03T params
  • 262,144 context
  • Released 2026-04-14(HF repo created)
15/27axes covered
Overall · LMArena
1,461#52 of 396
Best percentile
93.8IFBench (Artificial Analysis)
Cohort standing
69.2median 62.8

What the data says

Benchmarked on 15 of 27 tracked axes. Every figure below is the model's own published result on that board.

Strongest axes

  • 93.8 pct — IFBench (Artificial Analysis) (76.0%)
  • 90 pct — Coding (1,516)
  • 87.6 pct — Instruction following (1,456)

Uneven: 62.2 percentile points between its best and worst axis — 93.8 pct on IFBench (Artificial Analysis) versus 31.6 pct on Document Arena.

  • IFBench (Artificial Analysis)93.8 pct76.0%Verify →
  • Coding90 pct1,516Verify →
  • Instruction following87.6 pct1,456Verify →
  • Hard prompts87.6 pct1,486Verify →
  • Overall87.1 pct1,461Verify →
  • FrontierMath87 pct39.0%Verify →
  • GPQA Diamond84.6 pct90.8%Verify →
  • AIME (math)83 pct96.1%Verify →
  • Vision Arena78 pct1,265Verify →
  • SWE-bench Verified75 pct76.7%Verify →
  • Factuality Arena72.7 pct1,456Verify →
  • WebDev Arena64 pct1,510Verify →
  • SimpleQA Verified36.6 pct34.9%Verify →
  • HHEM Hallucination (Vectara)36.2 pct89.2%Verify →
  • Document Arena31.6 pct1,450Verify →
  • Search ArenaNot benchmarked
  • Agent ArenaNot benchmarked
  • ARC-AGI-2Not benchmarked
  • Humanity's Last ExamNot benchmarked
  • SimpleBenchNot benchmarked
  • LiveBenchNot benchmarked
  • BFCL v4 (function calling)Not benchmarked
  • SWE-bench Verified (live board)Not benchmarked
  • Terminal-Bench 4.0Not benchmarked
  • MMLU-Pro (Artificial Analysis)Not benchmarked
  • LiveCodeBench (Artificial Analysis)Not benchmarked
  • Terminal-BenchNot benchmarked

Percentiles are the model's standing within each board's own population. Axes marked "not benchmarked" have no published result for this model — that is an absence of data, not a low score.

How to run it

Run it yourself

Repositorymoonshotai/Kimi-K2.6
Parameters1.03T
ArchitectureKimiK25ForConditionalGeneration
LicenceModified MIT

Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "moonshotai/Kimi-K2.6"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id)

Ollama

We don't sync Ollama tags, so we won't print a command we can't verify — a wrong one would pull the wrong model. Search the Ollama library →

Unsloth fine-tuning recipes are model-specific and we don't derive them here. Unsloth docs →

Use it hosted

API model IDmoonshotai/kimi-k2.6
Context length262,144 tokens
Max output235,929 tokens
Acceptstext, image
Input price$0.65 per 1M tokens
Cached input price$0.15 per 1M tokens
Output price$3.41 per 1M tokens

API parameters this endpoint supports

Frequency penaltyInclude reasoningLogit biasLogprobsMax tokensMin-pParallel Tool CallsPresence penaltyReasoningRepetition penaltyResponse formatSeedStop sequencesStructured outputsTemperatureTool choiceTool callingTop-kTop logprobsTop-p

Example request

curl https://openrouter.ai/api/v1/chat/completions \
  -H "Authorization: Bearer $OPENROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"moonshotai/kimi-k2.6","messages":[{"role":"user","content":"Hello"}]}'

View on OpenRouter →

Identifiers, limits and prices are synced from the upstream source shown above. Any field we haven't synced is left out entirely rather than guessed.

  • kimi-k2.6 stands at 69.2 (raw mean 73) — measured on 15 of 27 axes.

    standingtemplateVerify →

  • kimi-k2.6 stands at 69.2 versus a cohort median of 62.8 across 146 models released within 180 days — above the peer median.

    cohorttemplateVerify →

Every claim is a computed fact from the current board data, phrased by a low-temperature model call and rejected by a validation gate if any number is not in the source row. How the leaderboards work →