Skip to content
AI Atlas
ModelActiveOpen weightsIdentity probable

Kimi K3

Moonshot AIfamily · Kimihuggingface.co/moonshotai/Kimi-K3

Kimi K3 is a 2.8T parameter open-weight multimodal reasoning model from Moonshot AI. It is suited for complex coding, knowledge work, and long-horizon agentic workflows, and is particularly strong at...

Open in Graph
data quality72

Updated 3 h ago · first seen 11 Sept 2026

model_01M294AHZM52W3RF44Y4H7K7JV

Overview

Identity

Canonical model
Yesidentity confidence: mediumOne row per real model release. Artifacts (checkpoints, quantisations, conversions) and folded evaluation variants point here.
Official checkpoints
official_checkpoints = hf_repo identifiers carried by the model itself; artifacts are separate entities pointing here through canonical_id.
Artifacts
None recordedSeparate entities (checkpoint · quantization · conversion · packaging) pointing to this model through canonical_id.
Provider deployments
4
API aliases
fireworks/kimi-k3kimi-k3kimi-k3-lowmoonshotai/kimi-k3Identifiers under which providers and evaluators refer to this model.
Folded evaluation variants
1Effort / thinking variants (…-high, …-non-reasoning) are result configurations of this model, not separate models. Their old URLs redirect here.

Openness

Open weights— weights downloadable under Other; 7 dimensions unknown.

Weights downloadable under a permissive or Creative Commons licence allowing commercial use; code or data may be missing.

  • Weights

    Yes

  • Inference code

    —

  • Training code

    —

  • Training data

    —

  • Dataset

    —

  • Commercial use

    —

  • Redistribution

    —

  • Derivatives

    —

Licence: Other (unclassified licence) (unknown · stated as “other”)

dimensions marked null are unknown, not false

Key facts

Release date

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 8 h agomedium

Model card
huggingface.co/moonshotai/Kimi-K3

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 13 h agomedium

Openrouter id

Source:OpenRouter public model & pricing listingT2observed 13 h agomedium

Architecture

Architecture

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 13 h agomedium

Model type

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 13 h agomedium

Parameters

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 13 h agomedium

Weights dtype

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 13 h agomedium

File size

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 13 h agomedium

Library name

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 13 h agomedium

Pipeline tag

Source:Hugging Face Hub (public pages, model cards, papers)T2observed 13 h agomedium

Hugging Face repo

Source:OpenRouter public model & pricing listingT2observed 13 h agomedium

Capabilities

Modalities

Modalities
imagetextvideo
Input
imagetextvideo
Output
text

Capabilities

  • Tool calling

    Yes

    OpenRouter public model & pricing listing · T2

  • Structured output

    Yes

    OpenRouter public model & pricing listing · T2

  • Reasoning

    Yes

    OpenRouter public model & pricing listing · T2

  • Vision

    Yes

    OpenRouter public model & pricing listing · T2

  • Audio

    Unavailable

  • Fine-tuning available

    Unavailable

Context window

Source:OpenRouter public model & pricing listingT2observed 13 h agomedium

Max output

Source:OpenRouter public model & pricing listingT2observed 13 h agomedium

Benchmarks35

Compare with another model →

Comparable same task and conditions · Partially comparable same task, conditions differ (effort, temperature, judge) · Not comparable different variant or metric

Benchmark results grouped by comparability group
Benchmark · groupBest scoreTrustConfigurationResultsvs leaderEvaluatedSource
Terminal-Benchagentic · accuracy · variant=v4.0 · evaluator=Artificial AnalysisIndependentevaluatorArtificial Analysisvariantv4.0reasoning_effortlowconditions differ across rows → partially comparable3non-primary groupobs. 12 Sept 2026artificialanalysis.aiT2
Terminal-Benchagentic · accuracy · variant=v2.1 · evaluator=Artificial AnalysisIndependentevaluatorArtificial Analysisvariantv2.1reasoning_effortmaxconditions differ across rows → partially comparable4non-primary groupobs. 12 Sept 2026artificialanalysis.aiT2
LiveBench Agentic Codingcoding · average score · variant=Agentic CodingOfficial boardvariantAgentic Codingrelease2026-06-251−15.1 ptvs DeepSeek-V4.1-Flash25 Jun 2026livebench.aiT2
LiveBench Codingcoding · average score · variant=CodingOfficial boardvariantCodingrelease2026-06-251−4.93 ptvs Claude Fable 5.125 Jun 2026livebench.aiT2
SciCodecoding · accuracy · evaluator=Artificial AnalysisIndependentevaluatorArtificial Analysisreasoning_effortmaxconditions differ across rows → partially comparable4−3.59 ptvs Claude Fable 5.1obs. 12 Sept 2026artificialanalysis.aiT2
Artificial Analysis Intelligence Indexcomposite · indexIndependentreasoning_effortmaxversion4.3conditions differ across rows → partially comparable4−9.59vs Claude Fable 5.1obs. 12 Sept 2026artificialanalysis.aiT2
LiveBenchgeneral · global_averageOfficial boardrelease2026-06-251−4.22 ptvs Claude Fable 5.125 Jun 2026livebench.aiT2
LiveBench Data Analysisgeneral · average score · variant=Data AnalysisOfficial boardvariantData Analysisrelease2026-06-251−4.24 ptvs gpt-6-astra25 Jun 2026livebench.aiT2
LiveBench Languagegeneral · average score · variant=LanguageOfficial boardvariantLanguagerelease2026-06-251−5.16 ptvs Claude Fable 525 Jun 2026livebench.aiT2
LiveBench Instruction Followinginstruction-following · average score · variant=IFOfficial boardvariantIFrelease2026-06-251−10.0 ptvs Gemini 3.8 Flash25 Jun 2026livebench.aiT2
Humanity's Last Examknowledge · accuracy · evaluator=Artificial AnalysisIndependentevaluatorArtificial Analysisreasoning_effortmaxconditions differ across rows → partially comparable4−12.2 ptvs Claude Fable 5.1obs. 12 Sept 2026artificialanalysis.aiT2
LiveBench Mathematicsmath · average score · variant=MathematicsOfficial boardvariantMathematicsrelease2026-06-251−12.6 ptvs Claude Fable 5.125 Jun 2026livebench.aiT2
MMMU-Promultimodal · accuracy · evaluator=Artificial AnalysisIndependentevaluatorArtificial Analysisreasoning_effortmaxconditions differ across rows → partially comparable4−6.36 ptvs gpt-6-astraobs. 12 Sept 2026artificialanalysis.aiT2
GPQA Diamondreasoning · accuracy · variant=Diamond · evaluator=Artificial AnalysisIndependentevaluatorArtificial AnalysisvariantDiamondreasoning_effortmaxconditions differ across rows → partially comparable2non-primary groupobs. 12 Sept 2026artificialanalysis.aiT2
GPQA Diamondreasoning · accuracy · variant=GPQA Diamond · evaluator=Artificial AnalysisIndependentevaluatorArtificial AnalysisvariantGPQA Diamond2−2.72 ptvs gpt-6-astraobs. 11 Sept 2026artificialanalysis.aiT2
LiveBench Reasoningreasoning · average score · variant=ReasoningOfficial boardvariantReasoningrelease2026-06-251−1.98 ptvs gpt-6-astra25 Jun 2026livebench.aiT2

Current rows only, grouped by benchmark → canonical metric → comparability group (task configuration). Effort variants folded into this model appear as rows of the same group. 35 current rows in total. “vs leader” compares with the current leader of the benchmark's primary group only; other groups are not directly comparable. Comparability rules →

Providers & Pricing8

All offers in the price terminal →
Provider deployments, cheapest output first
ProviderContextInput / 1MCached inOutput / 1MBatch in / outStatusObservedSource
Moonshot AI Platformcheapest outputmoonshotai/kimi-k31.05Mout 943.7K$0.263—active4 h agosince 12 Sept 2026openrouter.aiT2
OpenRoutermoonshotai/kimi-k31.05Mout 943.7K$0.263—active1 h agosince 12 Sept 2026openrouter.aiT2
Fireworks AIfireworks/kimi-k3—$0.30—active6 h agosince 11 Sept 2026app.fireworks.aiT1
Moonshot AI Platformmoonshotai/kimi-k3:batch1.05Mout 943.7K$0.30—active4 h agosince 11 Sept 2026openrouter.aiT2
OpenRoutermoonshotai/kimi-k3:batch1.05Mout 943.7K$0.30—active1 h agosince 12 Sept 2026openrouter.aiT2
Together AIkimi-k3—$0.30$3 / $15active12 h agosince 11 Sept 2026together.aiT1
Fireworks AIfireworks/kimi-k3:us—$0.33—active6 h agosince 11 Sept 2026app.fireworks.aiT1
Fireworks AIfireworks/kimi-k3:fast—$0.45—active6 h agosince 11 Sept 2026app.fireworks.aiT1

USD per 1M tokens as published by each provider; native units (per-request fees, flex/priority tiers) are kept verbatim. Rows are append-only — every price change is kept in the history below. Cost of a workload →

Price history

Step lines per provider; amber markers are recorded changes. Click a marker or a row for the evidence behind that price.

Output price · USD / 1M tokens 4 providers

Output price history of Kimi K3$0$10$20$30Sept 26Sept 26Sept 26Sept 26Fireworks AI: first observed → $15 · 11 Sept 2026Fireworks AI: $15 → $22.5 · 11 Sept 2026Fireworks AI: $22.5 → $16.5 · 11 Sept 2026Moonshot AI Platform: first observed → $9.01 · 11 Sept 2026Moonshot AI Platform: $9.01 → $15 · 11 Sept 2026Together AI: first observed → $15 · 11 Sept 2026Moonshot AI Platform: $15 → $7.7 · 12 Sept 2026Moonshot AI Platform: $7.7 → $11.55 · 12 Sept 2026OpenRouter: first observed → $11.55 · 12 Sept 2026OpenRouter: $11.55 → $15 · 12 Sept 2026
  • Fireworks AI
  • Moonshot AI Platform
  • Together AI
  • OpenRouter
  • OpenRouter$11.55 → $1512 Sept 2026
  • OpenRouterfirst observed $11.5512 Sept 2026
  • Moonshot AI Platform$7.7 → $11.5512 Sept 2026
  • Moonshot AI Platform$15 → $7.712 Sept 2026
  • Together AIfirst observed $1511 Sept 2026
  • Moonshot AI Platform$9.01 → $1511 Sept 2026
  • Moonshot AI Platformfirst observed $9.0111 Sept 2026
  • Fireworks AI$22.5 → $16.511 Sept 2026
  • Fireworks AI$15 → $22.511 Sept 2026
  • Fireworks AIfirst observed $1511 Sept 2026

Input price · USD / 1M tokens 4 providers

Input price history of Kimi K3$0$2$4$6Sept 26Sept 26Sept 26Sept 26Fireworks AI: first observed → $3 · 11 Sept 2026Fireworks AI: $3 → $4.5 · 11 Sept 2026Fireworks AI: $4.5 → $3.3 · 11 Sept 2026Moonshot AI Platform: first observed → $1.8 · 11 Sept 2026Moonshot AI Platform: $1.8 → $3 · 11 Sept 2026Together AI: first observed → $3 · 11 Sept 2026Moonshot AI Platform: $3 → $1.54 · 12 Sept 2026Moonshot AI Platform: $1.54 → $2.3 · 12 Sept 2026OpenRouter: first observed → $2.3 · 12 Sept 2026OpenRouter: $2.3 → $3 · 12 Sept 2026
  • Fireworks AI
  • Moonshot AI Platform
  • Together AI
  • OpenRouter
  • OpenRouter$2.3 → $312 Sept 2026
  • OpenRouterfirst observed $2.312 Sept 2026
  • Moonshot AI Platform$1.54 → $2.312 Sept 2026
  • Moonshot AI Platform$3 → $1.5412 Sept 2026
  • Together AIfirst observed $311 Sept 2026
  • Moonshot AI Platform$1.8 → $311 Sept 2026
  • Moonshot AI Platformfirst observed $1.811 Sept 2026
  • Fireworks AI$4.5 → $3.311 Sept 2026
  • Fireworks AI$3 → $4.511 Sept 2026
  • Fireworks AIfirst observed $311 Sept 2026

Hardware fit37

Estimated

0 of 37 device × quantization combinations fit.

Run locally: your machine →
Estimated hardware fit
HardwareQuantizationDevice memoryEst. memoryFits
Apple M3 Ultra4bit—1599.0 GB est.No
Mac Studio (Apple M5 Ultra)4bit—1599.0 GB est.No
Apple M2 Ultra4bit—1599.0 GB est.No
Apple M1 Ultra4bit—1599.0 GB est.No
Apple M3 Max4bit—1599.0 GB est.No
Apple M4 Max4bit—1599.0 GB est.No
Mac Studio (Apple M5 Max)4bit—1599.0 GB est.No
MacBook Pro (Apple M5 Max)4bit—1599.0 GB est.No
Apple M2 Max4bit—1599.0 GB est.No
Apple M1 Max4bit—1599.0 GB est.No
Apple M4 Pro4bit—1599.0 GB est.No
Mac mini (Apple M5 Pro)4bit—1599.0 GB est.No
MacBook Pro (Apple M5 Pro)4bit—1599.0 GB est.No
Apple M3 Pro4bit—1599.0 GB est.No
Apple M1 Pro4bit—1599.0 GB est.No
Apple M2 Pro4bit—1599.0 GB est.No
Apple M44bit—1599.0 GB est.No
iMac (Apple M4)4bit—1599.0 GB est.No
Mac mini (Apple M6)4bit—1599.0 GB est.No
MacBook Air (Apple M5)4bit—1599.0 GB est.No
MacBook Pro (Apple M5)4bit—1599.0 GB est.No
Apple M24bit—1599.0 GB est.No
Apple M34bit—1599.0 GB est.No
Apple M14bit—1599.0 GB est.No
NVIDIA GeForce RTX 30904bit24 GB1599.0 GB est.No
NVIDIA GeForce RTX 40904bit24 GB1599.0 GB est.No
NVIDIA GeForce RTX 50904bit32 GB1599.0 GB est.No
NVIDIA A100 80GB4bit80 GB1599.0 GB est.No
NVIDIA H100 SXM4bit80 GB1599.0 GB est.No
NVIDIA H100 NVL4bit94 GB1599.0 GB est.No
NVIDIA DGX Spark4bit128 GB1599.0 GB est.No
NVIDIA H2004bit141 GB1599.0 GB est.No
NVIDIA H200 NVL4bit141 GB1599.0 GB est.No
NVIDIA B2004bit180 GB1599.0 GB est.No
AMD Instinct MI300X4bit192 GB1599.0 GB est.No
AMD Instinct MI325X4bit256 GB1599.0 GB est.No
NVIDIA DGX B2004bit1,440 GB1599.0 GB est.No
Assumptions (7)
  • Estimated, not measured: weights = parameters × bytes/param × 1.15 runtime overhead (or the observed artifact file size when one is recorded).
  • bytes/param: 4bit = 0.5, 8bit = 1.0, fp16 = 2.0 (uniform quantization, no per-layer exceptions).
  • KV cache: 2 × layers × kv_heads × head_dim × 2 bytes × context × batch when the architecture is known; otherwise 0.5 GB per 8 192 tokens (× batch), independent of architecture (GQA/MLA models need less).
  • A model 'fits' when the estimate is at most the device memory minus 2 GB reserved for the OS and framework.
  • Mixture-of-experts models are estimated on total parameters (all experts must be resident); active parameters are ignored.
  • Device memory uses the largest configuration when several are listed (e.g. Apple silicon tiers).
  • Multi-GPU: device memories are summed; interconnect bandwidth, tensor-parallel replication and pipeline bubbles are not modelled.

Versions & Artifacts0

Version history

Context windowfirst observation only

11 Sept 2026current

Licensefirst observation only

11 Sept 2026current

Max outputfirst observation only

11 Sept 2026current

Opennessfirst observation only

11 Sept 2026current

Parametersfirst observation only

11 Sept 2026current

Each hop is a claim: click a value for its source, tier and observation time. Nothing is overwritten — a new observation closes the previous claim.

Artifacts 0

No artifact (checkpoint, quantisation, conversion or packaging) points to this model yet.

Timeline20

Full timeline →
  • Listed by providerModelKimi K3Moonshot AI

    OpenRouter lists Kimi K3 at $3 in / $15 out per 1M tokens

    openrouter
    openrouter.ai
  • Listed by providerModelKimi K3Moonshot AI

    OpenRouter lists Kimi K3 at $2.30273 in / $11.5502 out per 1M tokens

    openrouter
    openrouter.ai
  • Benchmark resultModelKimi K3Moonshot AI

    Kimi K3 scores 82.4% on Terminal-Bench

    artificial_analysis
    artificialanalysis.ai
  • Benchmark resultModelKimi K3Moonshot AI

    Kimi K3 scores 12.63% on Terminal-Bench

    artificial_analysis
    artificialanalysis.ai
  • Benchmark resultModelKimi K3Moonshot AI

    Kimi K3 scores 78.5% on MMMU-Pro

    artificial_analysis
    artificialanalysis.ai
  • Benchmark resultModelKimi K3Moonshot AI

    Kimi K3 scores 52.66% on SciCode

    artificial_analysis
    artificialanalysis.ai
  • Benchmark resultModelKimi K3Moonshot AI

    Kimi K3 scores 24.98% on Humanity's Last Exam

    artificial_analysis
    artificialanalysis.ai
  • Benchmark resultModelKimi K3Moonshot AI

    Kimi K3 scores 84.24% on GPQA Diamond

    artificial_analysis
    artificialanalysis.ai
  • Benchmark resultModelKimi K3Moonshot AI

    Kimi K3 scores 30.48 on Artificial Analysis Intelligence Index

    artificial_analysis
    artificialanalysis.ai
  • Benchmark resultModelKimi K3Moonshot AI

    Kimi K3 scores 85.02% on Terminal-Bench

    artificial_analysis
    artificialanalysis.ai
  • Benchmark resultModelKimi K3Moonshot AI

    Kimi K3 scores 12.63% on Terminal-Bench

    artificial_analysis
    artificialanalysis.ai
  • Benchmark resultModelKimi K3Moonshot AI

    Kimi K3 scores 80.52% on MMMU-Pro

    artificial_analysis
    artificialanalysis.ai
  • Benchmark resultModelKimi K3Moonshot AI

    Kimi K3 scores 59.49% on SciCode

    artificial_analysis
    artificialanalysis.ai
  • Benchmark resultModelKimi K3Moonshot AI

    Kimi K3 scores 46.9% on Humanity's Last Exam

    artificial_analysis
    artificialanalysis.ai
  • Benchmark resultModelKimi K3Moonshot AI

    Kimi K3 scores 93.54% on GPQA Diamond

    artificial_analysis
    artificialanalysis.ai
  • Benchmark resultModelKimi K3Moonshot AI

    Kimi K3 scores 43.78 on Artificial Analysis Intelligence Index

    artificial_analysis
    artificialanalysis.ai
  • Price changedModelKimi K3Moonshot AI

    Moonshot AI Platform changed pricing for Kimi K3: $1.53515 in / $7.70013 out per 1M tokens → $2.30273 in / $11.5502 out per 1M tokens

    $1.54 in / $7.7 out→$2.3 in / $11.55 outopenrouter
    openrouter.ai
  • Price changedModelKimi K3Moonshot AI

    Moonshot AI Platform changed pricing for Kimi K3: $1.7955 in / $9.006 out per 1M tokens → $1.53515 in / $7.70013 out per 1M tokens

    $1.8 in / $9.01 out→$1.54 in / $7.7 outopenrouter
    openrouter.ai
  • Release date changedModelKimi K3Moonshot AI

    Kimi K3: release date changed from 2026-06-13 to 2026-07-16

    Release date13 Jun 2026→16 Jul 2026openrouter
    openrouter.ai
  • Release date changedModelKimi K3Moonshot AI

    Kimi K3: release date changed from 2026-07-16 to 2026-06-13

    Release date16 Jul 2026→13 Jun 2026huggingface
    huggingface.co

Change history

Temporal, append-only claims: a new observation closes the previous claim instead of overwriting it. Rewind the record with the as-of picker.
1 claims · 1 propertiesShow all properties

Model typemodel_type1

Claim history for Model type
ValueValid from → toStatusSourceConfidenceExtractor
kimi_k3 → currentcurrentHugging Face Hub (public pages, model cards, papers)T2mediumdeterministic

Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →

Provenance

Attributed facts

40

Source tiers

T240

Freshest observation

3 h ago

Conflicts

None

Source documents 8

Source documents
SourceDocumentTypeTierLast observedSnapshots
Together AI — pricingtogether.ai/pricing pricingT1· Official2 h ago1
Fireworks AI — pricingdocs.fireworks.ai/serverless/pricing.md pricingT1· Official3 h ago2
OpenRouter public model & pricing listingopenrouter.ai/api/v1/models catalogueT2· Quality secondary1 h ago10
Artificial Analysisartificialanalysis.ai/leaderboards/models leaderboardT2· Quality secondary3 h ago2
Hugging Face Hub (public pages, model cards, papers)huggingface.co/models?author=moonshotai&p=0&sort=downloads listingT2· Quality secondary3 h ago7
Hugging Face Hub (public pages, model cards, papers)huggingface.co/moonshotai/Kimi-K3/raw/main/README.md model_cardT2· Quality secondary7 h ago1
Hugging Face Hub (public pages, model cards, papers)huggingface.co/moonshotai/Kimi-K3 model_pageT2· Quality secondary7 h ago5
LiveBenchlivebench.ai/table_2026_06_25.csv leaderboardT2· Quality secondary12 h ago1

Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.

Data quality (72/100) measures how well AI Atlas knows this entity — completeness, primary-source ratio, freshness, conflicts — never how good the model is. Methodology →