Skip to content
aiqbench

Moonshot · model passport

MoonshotAI: Kimi K2.5

kimi-k2-5
  • Open weights
  • Modified MIT licence
  • 1.03T params
  • 262,144 context
  • Released 2026-01-01(HF repo created)
9/27axes covered
Best percentile
82.1IFBench (Artificial Analysis)
Cohort standing
49.4median 55.5

What the data says

Benchmarked on 9 of 27 tracked axes. Every figure below is the model's own published result on that board.

Uneven: 65.9 percentile points between its best and worst axis — 82.1 pct on IFBench (Artificial Analysis) versus 16.2 pct on HHEM Hallucination (Vectara).

  • IFBench (Artificial Analysis)82.1 pct70.2%Verify →
  • Humanity's Last Exam65.2 pct24.4%Verify →
  • Terminal-Bench52.3 pct43.2%Verify →
  • SWE-bench Verified50 pct73.8%Verify →
  • WebDev Arena49.6 pct1,437Verify →
  • ARC-AGI-246.1 pct11.8%Verify →
  • SimpleBench46 pct46.8%Verify →
  • SimpleQA Verified35.4 pct34.3%Verify →
  • HHEM Hallucination (Vectara)16.2 pct85.8%Verify →
  • OverallNot benchmarked
  • Vision ArenaNot benchmarked
  • Factuality ArenaNot benchmarked
  • Search ArenaNot benchmarked
  • Document ArenaNot benchmarked
  • Agent ArenaNot benchmarked
  • GPQA DiamondNot benchmarked
  • AIME (math)Not benchmarked
  • FrontierMathNot 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
  • CodingNot benchmarked
  • Instruction followingNot benchmarked
  • Hard promptsNot 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-K2.5
Parameters1.03T
ArchitectureKimiK25ForConditionalGeneration
LicenceModified MIT

Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "moonshotai/Kimi-K2.5"
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-k2.5
Context length262,144 tokens
Max output235,929 tokens
Acceptstext, image
Input price$0.50 per 1M tokens
Cached input price$0.15 per 1M tokens
Output price$2.50 per 1M tokens

API parameters this endpoint supports

Frequency penaltyInclude reasoningLogit biasLogprobsMax tokensMin-pPresence 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-k2.5","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 K2.5 (Thinking) stands at 49.4 (raw mean 49.2) — measured on 9 of 27 axes.

    standingtemplateVerify →

  • Kimi K2.5 (Thinking) stands at 49.4 versus a cohort median of 55.5 across 120 models released within 180 days — below 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 →