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 navigating large repositories, using tools, debugging, and iterating against images, logs, tests, and runtime feedback. Its architecture uses KDA and Attention Residuals for computational efficiency.

moonshotai/kimi-k3
Organization
Moonshot AI
Family
kimi-k3
Providers
82
Context
1,048,576
Output limit
131,072
Knowledge
—
Release
2026-07-16
Updated
—
Weights
Open
Input
text, image/jpeg, image/png, image/gif, image/webp, image/bmp, image/heic, image/heif, video/mp4, video/mpeg, video/mov, video/avi, video/x-flv, video/mpg, video/webm, video/wmv, video/3gpp
Output
text
Capabilities
—

Providers

Compare every cataloged offer for this model, including its provider, route, token limits, price, and capabilities.

133 offers · USD per million tokens
ProviderModel IDContextOutputInput / output · 1MReasoningToolsStructuredDetails
302.AI
kimi-k3
1,048,576131,072— / ————
View
Abacus
moonshotai/Kimi-K3
1,048,576131,072— / ————
View
ai&
moonshotai/kimi-k3
1,048,576131,072— / ————
View
AIHubMix
kimi-k3
1,048,576131,072— / ————
View
AI/ML API
moonshot/kimi-k3
1,048,576131,072— / ————
View
ainetcafe
Kimi-K3
262,14432,768— / ————
View
Alibaba
kimi-k3
1,048,5761,048,576— / ————
View
Alibaba (China)
kimi-k3
1,048,5761,048,576— / ————
View
Amazon Bedrock
us.moonshotai.kimi-k3
1,048,576128,000— / ————
View
Amazon Bedrock
global.moonshotai.kimi-k3
1,048,576128,000— / ————
View
Arcee
moonshotai/kimi-k3
1,048,576131,072— / ————
View
Baseten
moonshotai/Kimi-K3
1,048,576262,144— / ————
View
Berget.AI
moonshotai/Kimi-K3
327,68032,768— / ————
View
Boundless
kimi-k3
1,048,576131,072— / ————
View
Charm Hyper
kimi-k3
1,048,57616,000— / ————
View
Chutes
moonshotai/Kimi-K3-TEE
1,048,57665,535— / ————
View
ClinePass
cline-pass/kimi-k3
1,048,576131,072— / ————
View
Cloudflare AI Gateway
moonshotai/kimi-k3
1,048,576131,072— / ————
View
CommandCode
moonshotai/Kimi-K3
1,000,000131,072— / ————
View
Cortecs
kimi-k3
1,048,5761,048,576— / ————
View
CrossModel
moonshot/kimi-k3
1,048,5761,048,576— / ————
View
Deep Infra
moonshotai/Kimi-K3
1,048,576131,072— / ————
View
DevPass (LLM Gateway)
kimi-k3
1,048,576131,072— / ————
View
DevPass (LLM Gateway)
kimi-k3-fast
1,040,384131,072— / ————
View
DigitalOcean
kimi-k3
1,048,576131,072— / ————
View
Eden AI
moonshot/kimi-k3
1,048,576131,072— / ————
View
EmpirioLabs AI
kimi-k3
1,000,000131,072— / ————
View
engy
kimi-k3
1,113,08865,536— / ————
View
Fireworks AI
accounts/fireworks/models/kimi-k3
1,048,576131,072— / ————
View
Fireworks AI
accounts/fireworks/routers/kimi-k3-fast
1,048,576131,072— / ————
View
Fireworks AI
accounts/fireworks/routers/kimi-k3-spot
1,048,576131,072— / ————
View
Fireworks AI
accounts/fireworks/routers/kimi-latest
1,048,576131,072— / ————
View
Fireworks AI
accounts/fireworks/routers/kimi-fast-latest
1,048,576131,072— / ————
View
GitHub Copilot
kimi-k3
1,048,576131,072— / ————
View
GreenPT
kimi-k3
1,048,576131,072— / ————
View
Hugging Face
moonshotai/Kimi-K3
1,000,000131,072— / ————
View
Impossibl
moonshotai/kimi-k3
1,048,576131,072— / ————
View
Inco
kimi-k3:fast
1,048,576131,072— / ————
View
IteraCompute
moonshotai/kimi-k3
1,048,576999,999— / ————
View
Jalapeno Cloud
Kimi-K3
1,048,576131,072— / ————
View
Kenari
kimi-k3
1,048,576131,072— / ————
View
Kilo Gateway
moonshotai/kimi-k3
1,048,576943,718— / ————
View
Kimi For Coding (kimi.ai)
k3-256k
262,144131,072— / ————
View
Kimi For Coding (kimi.ai)
k3
1,048,576131,072— / ————
View
Kimi For Coding (kimi.com)
k3-256k
262,144131,072— / ————
View
Kimi For Coding (kimi.com)
k3
1,048,576131,072— / ————
View
LLM Gateway
aws-bedrock/kimi-k3
1,048,5761,048,576— / ————
View
LLM Gateway
together-ai/kimi-k3
1,000,0001,000,000— / ————
View
LLM Gateway
novita/kimi-k3
1,048,5761,048,576— / ————
View
LLM Gateway
scx-ai-gp/kimi-k3
1,048,5761,048,576— / ————
View
LLM Gateway
tencent/kimi-k3
1,048,5761,048,576— / ————
View
LLM Gateway
fireworks/kimi-k3
1,040,3841,040,384— / ————
View
LLM Gateway
canopywave/kimi-k3
1,048,5761,048,576— / ————
View
LLM Gateway
runpod/kimi-k3
1,048,5761,048,576— / ————
View
LLM Gateway
runware/kimi-k3
1,048,5761,048,576— / ————
View
LLM Gateway
moonshot/kimi-k3
1,048,5761,048,576— / ————
View
LLM Gateway
alibaba/kimi-k3
1,048,5761,048,576— / ————
View
LLM Gateway
fireworks/kimi-k3-fast
1,040,3841,040,384— / ————
View
Melious
kimi-k3
1,000,000131,072— / ————
View
Merge Gateway
moonshot/kimi-k3
1,048,5761,048,576— / ————
View
Moark
kimi-k3
1,048,576131,072— / ————
View
Modal
moonshotai/Kimi-K3
1,048,576131,072— / ————
View
Moonshot AI
kimi-k3
1,048,5761,048,576— / ————
View
Moonshot AI (China)
kimi-k3
1,048,5761,048,576— / ————
View
NanoGPT
TEE/kimi-k3
1,048,57665,535— / ————
View
NanoGPT
moonshotai/kimi-k3
1,048,576943,718— / ————
View
NEAR AI Cloud
moonshotai/kimi-k3
1,048,576131,072— / ————
View
Nebius Token Factory
moonshotai/Kimi-K3
1,024,0008,000— / ————
View
Neon
kimi-k3
1,048,57665,536— / ————
View
Neuralwatt
kimi-k3-flex
1,048,5601,048,560— / ————
View
Neuralwatt
kimi-k3
1,048,5601,048,560— / ————
View
Neuralwatt
kimi-k3-fast
1,048,5601,048,560— / ————
View
Novita AI
moonshotai/kimi-k3
1,048,5761,048,576— / ————
View
Nvidia
moonshotai/kimi-k3
1,048,576131,072— / ————
View
Ofox
moonshotai/kimi-k3
1,048,576131,072— / ————
View
Ollama Cloud
kimi-k3
1,048,576131,072— / ————
View
OpenCode
kimi-k3
1,048,576131,072— / ————
View
OpenCode Go
kimi-k3
1,048,576131,072— / ————
View
AkashMLvia OpenRouter
moonshotai/kimi-k3
fp4
1,048,576943,718$1.20 / $14.00YesYesYes
View
Alibabavia OpenRouter
moonshotai/kimi-k3
unknown
1,048,576131,072$3.45 / $17.25YesYesYes
View
Amazon Bedrockvia OpenRouter
moonshotai/kimi-k3
unknown
1,048,576128,000$3.00 / $15.00YesYesNo
View
BaseTenvia OpenRouter
moonshotai/kimi-k3
fp8
1,048,576262,144$3.00 / $15.00YesYesNo
View
Chutesvia OpenRouter
moonshotai/kimi-k3
mxfp4
1,048,57665,535$3.00 / $15.00YesNoYes
View
Decartvia OpenRouter
moonshotai/kimi-k3
mxfp4
1,048,576943,718$1.98 / $9.90YesYesYes
View
DeepInfravia OpenRouter
moonshotai/kimi-k3
mxfp4
1,048,576943,718$2.85 / $14.25YesYesYes
View
DigitalOceanvia OpenRouter
moonshotai/kimi-k3
unknown
1,048,576943,718$2.55 / $12.95YesYesYes
View
Fireworksvia OpenRouter
moonshotai/kimi-k3
unknown
1,048,576943,718$4.50 / $22.50YesYesYes
View
Fireworksvia OpenRouter
moonshotai/kimi-k3
unknown
1,048,576943,718$4.50 / $22.50YesYesYes
View
Fireworksvia OpenRouter
moonshotai/kimi-k3
unknown
1,048,576943,718$3.00 / $15.00YesYesYes
View
InferenceNetvia OpenRouter
moonshotai/kimi-k3
fp4
250,000225,000$2.99 / $15.00YesYesYes
View
InferenceNetvia OpenRouter
moonshotai/kimi-k3
fp4
1,048,576943,718$0.64 / $13.50YesYesYes
View
Makoravia OpenRouter
moonshotai/kimi-k3
fp4
1,048,576943,718$1.53 / $12.75YesYesYes
View
Modalvia OpenRouter
moonshotai/kimi-k3
mxfp4
1,048,576943,718$3.00 / $15.00YesYesYes
View
Moonshot AIvia OpenRouter
moonshotai/kimi-k3
mxfp4
1,048,576943,718$3.00 / $15.00YesYesYes
View
Morphvia OpenRouter
moonshotai/kimi-k3
fp8
1,048,576943,718$0.339 / $14.90YesYesYes
View
Parasailvia OpenRouter
moonshotai/kimi-k3
fp4
1,048,576943,718$2.60 / $13.00YesYesYes
View
Parasailvia OpenRouter
moonshotai/kimi-k3
fp4
1,000,000900,000$3.00 / $15.00YesNoYes
View
Phalavia OpenRouter
moonshotai/kimi-k3
unknown
1,048,576943,718$2.55 / $12.75YesYesYes
View
Relacevia OpenRouter
moonshotai/kimi-k3
fp4
1,048,576943,718$2.00 / $14.00YesYesNo
View
Sail Researchvia OpenRouter
moonshotai/kimi-k3
fp4
1,048,576943,718$0.84 / $13.50YesYesYes
View
Togethervia OpenRouter
moonshotai/kimi-k3
unknown
1,048,576943,718$2.70 / $13.50YesYesYes
View
Wafervia OpenRouter
moonshotai/kimi-k3
unknown
1,048,576943,718$0.33 / $14.89YesYesYes
View
Wafervia OpenRouter
moonshotai/kimi-k3
unknown
1,048,576943,718$2.80 / $14.00YesYesYes
View
Opper
kimi-k3
1,048,576131,072— / ————
View
OrcaRouter
kimi/kimi-k3
1,048,576131,072— / ————
View
Perplexity Agent
perplexity/kimi-k3
1,048,576131,072— / ————
View
Perplexity Agent
moonshot-ai/kimi-k3
1,048,576131,072— / ————
View
Pioneer
moonshotai/Kimi-K3
1,000,000131,072— / ————
View
Pioneer
moonshotai/Kimi-K3-Fast
1,000,000131,072— / ————
View
Prime Inference
moonshotai/kimi-k3
1,048,576943,718— / ————
View
Requesty
kimi-k3@eu
1,048,576262,144— / ————
View
Requesty
kimi-k3
1,048,576262,144— / ————
View
SCNet Token Plan
Kimi-K3
1,048,576131,072— / ————
View
SenseNova (China)
kimi-k3
1,048,57665,536— / ————
View
SiliconFlow
moonshotai/Kimi-K3
1,048,576262,000— / ————
View
Standard Compute
moonshotai/kimi-k3
1,048,57650,000— / ————
View
Synthetic
hf:moonshotai/Kimi-K3
524,28865,536— / ————
View
Tempr Gateway
moonshot/kimi-k3
1,048,576131,072— / ————
View
TensorX
moonshotai/kimi-k3
1,048,576131,072— / ————
View
Tinfoil
kimi-k3
262,144131,072— / ————
View
Together AI
moonshotai/Kimi-K3
1,048,576131,072— / ————
View
TokenGo
moonshotai/kimi-k3
1,048,576131,072— / ————
View
Umans AI
umans-kimi-k3
1,048,576131,072— / ————
View
Umans AI Coding Plan
umans-kimi-k3
1,048,576131,072— / ————
View
Vancine
kimi-k3
1,048,576131,072— / ————
View
Venice AI
kimi-k3
1,000,000131,072— / ————
View
Venice AI
kimi-k3-fast-api
1,000,000131,072— / ————
View
Vercel AI Gateway
moonshotai/kimi-k3
1,000,000131,072— / ————
View
Vercel AI Gateway
moonshotai/kimi-k3-fast
1,000,000131,072— / ————
View
Vivgrid
kimi-k3
1,000,000131,072— / ————
View
Volcengine Ark Coding Plan
kimi-k3
1,048,576131,072— / ————
View
Wallaby
moonshotai/kimi-k3
1,048,576131,072— / ————
View
ZenMux
moonshotai/kimi-k3
1,048,576131,072— / ————
View

Prices and limits apply to each offer. A dash means the value is not provided. OpenRouter offers show the hosting provider and routing channel separately.

Pricing

Compare token rates and cache charges across available routes and hosting configurations.

USD per million tokens
Input from$0.33Per 1M input tokens
Output from$9.90Per 1M output tokens
Available offers133Across 82 providers

Lowest listed input and output prices across the offers above; they may belong to different providers. These are base rates, before conditional pricing, discounts or additional fees.

Endpoint price comparisonLatest observed snapshot · 10 lowest-priced configurations
InputOutput
Waferwafer
$0.33$14.89
Morphmorph/fp8
$0.339$14.90
InferenceNetinference-net/fp4
$0.64$13.50
Sail Researchsail-research/fp4
$0.84$13.50
AkashMLakashml/fp4
$1.20$14.00
Makoramakora/fp4
$1.53$12.75
Decartdecart/mxfp4
$1.98$9.90
Relacerelace/fp4
$2.00$14.00
DigitalOceandigitalocean
$2.55$12.95
Phalaphala
$2.55$12.75
moonshotai/kimi-k3Observed
Input
$0.64
Output
$13.50
Cache read
$0.28
Hosting prices through OpenRouter All endpoints and cache rates
EndpointInput / 1MOutput / 1MCache read / 1MCache write / 1M
AkashMLakashml/fp4$1.20$14.00$1.20—
Alibabaalibaba$3.45$17.25$0.345—
Amazon Bedrockamazon-bedrock/us-east-2$3.00$15.00$0.30$3.75
BaseTenbaseten/fp8$3.00$15.00$0.30—
Chuteschutes/mxfp4$3.00$15.00$0.30—
Decartdecart/mxfp4$1.98$9.90$0.198—
DeepInfradeepinfra/mxfp4$2.85$14.25$0.285—
DigitalOceandigitalocean$2.55$12.95$0.255—
Fireworksfireworks/us$4.50$22.50$0.45—
Fireworksfireworks/fast$4.50$22.50$0.45—
Fireworksfireworks$3.00$15.00$0.30—
InferenceNetinference-net/fast$2.99$15.00$0.45—
InferenceNetinference-net/fp4$0.64$13.50$0.28—
Makoramakora/fp4$1.53$12.75$0.204—
Modalmodal/mxfp4$3.00$15.00$0.30—
Moonshot AImoonshotai/mxfp4$3.00$15.00$0.30—
Morphmorph/fp8$0.339$14.90$0.29—
Parasailparasail/fp4$2.60$13.00$0.26—
Parasailparasail/fast$3.00$15.00$0.30—
Phalaphala$2.55$12.75$0.255—
Relacerelace/fp4$2.00$14.00$0.30—
Sail Researchsail-research/fp4$0.84$13.50$0.30—
Togethertogether$2.70$13.50$0.27—
Waferwafer$0.33$14.89$0.30—
Waferwafer/us$2.80$14.00$0.30—

Each row is a hosting configuration. Open an offer in Providers for its full pricing conditions and additional fees. Endpoint observations may differ from the routing catalog's base price.

Performance

Review reported latency and output-throughput percentiles from the latest observed rolling window.

Latest 30-minute observation

Time to first token · milliseconds

Latency distributionRolling 30-minute snapshot · 6 endpoints
P50P75P90P99
AkashMLakashml/fp4
1,202.5–2,526.97 ms
Alibabaalibaba
1,811–7,935.19 ms
Amazon Bedrockamazon-bedrock/us-east-2
1,021–3,976.32 ms
BaseTenbaseten/fp8
4,510–19,793.8 ms
Chuteschutes/mxfp4
2,011–8,439.96 ms
Decartdecart/mxfp4
2,501–23,740.59 ms
All latency percentiles 25 endpoints
EndpointP50P75P90P99
AkashMLakashml/fp41,202.5 ms1,386.75 ms1,525.9 ms2,526.97 ms
Alibabaalibaba1,811 ms2,335 ms3,124.4 ms7,935.19 ms
Amazon Bedrockamazon-bedrock/us-east-21,021 ms1,249.5 ms1,656.3 ms3,976.32 ms
BaseTenbaseten/fp84,510 ms6,458.75 ms8,998.5 ms19,793.8 ms
Chuteschutes/mxfp42,011 ms2,396 ms4,184 ms8,439.96 ms
Decartdecart/mxfp42,501 ms10,002.5 ms16,042.3 ms23,740.59 ms
DeepInfradeepinfra/mxfp41,858.5 ms2,888.75 ms5,489.7 ms16,009.35 ms
DigitalOceandigitalocean2,364 ms3,200 ms4,129.6 ms11,511.34 ms
Fireworksfireworks/us1,212.5 ms1,582.5 ms2,292.5 ms5,933.5 ms
Fireworksfireworks/fast1,114 ms2,330.5 ms4,301 ms13,475.39 ms
Fireworksfireworks1,019 ms1,612 ms3,055 ms5,197.26 ms
InferenceNetinference-net/fast833 ms1,256 ms1,798 ms4,535.28 ms
InferenceNetinference-net/fp4893 ms1,422 ms2,237.8 ms8,562.42 ms
Makoramakora/fp41,820.5 ms3,070.75 ms6,500.1 ms48,950.79 ms
Modalmodal/mxfp41,241 ms2,117 ms3,487.2 ms13,218.04 ms
Moonshot AImoonshotai/mxfp44,182 ms5,319 ms7,266.9 ms33,236.95 ms
Morphmorph/fp81,267 ms2,277.5 ms3,204.8 ms6,445.56 ms
Parasailparasail/fp41,232.5 ms1,934.75 ms3,191.8 ms11,077.1 ms
Parasailparasail/fast974 ms1,433 ms1,788.4 ms9,759.36 ms
Phalaphala1,450 ms2,423 ms4,108.6 ms12,515.5 ms
Relacerelace/fp41,518 ms2,489.5 ms6,403.2 ms29,965.6 ms
Sail Researchsail-research/fp42,518 ms3,812 ms7,600 ms19,890 ms
Togethertogether1,072 ms1,459 ms2,015 ms5,948.12 ms
Waferwafer1,015 ms1,749.75 ms3,886 ms21,463.49 ms
Waferwafer/us738 ms1,380 ms2,609 ms13,866.2 ms

Output throughput · tokens per second

Throughput distributionRolling 30-minute snapshot · 6 endpoints
P50P75P90P99
AkashMLakashml/fp4
17–29.87 t/s
Alibabaalibaba
38–68 t/s
Amazon Bedrockamazon-bedrock/us-east-2
72–141 t/s
BaseTenbaseten/fp8
48–120.08 t/s
Chuteschutes/mxfp4
26–34 t/s
Decartdecart/mxfp4
22–81.54 t/s
All throughput percentiles 25 endpoints
EndpointP50P75P90P99
AkashMLakashml/fp417 t/s20 t/s24 t/s29.87 t/s
Alibabaalibaba38 t/s44 t/s50 t/s68 t/s
Amazon Bedrockamazon-bedrock/us-east-272 t/s88 t/s101 t/s141 t/s
BaseTenbaseten/fp848 t/s71 t/s91 t/s120.08 t/s
Chuteschutes/mxfp426 t/s32 t/s33 t/s34 t/s
Decartdecart/mxfp422 t/s32 t/s45 t/s81.54 t/s
DeepInfradeepinfra/mxfp440 t/s48 t/s58 t/s79 t/s
DigitalOceandigitalocean34 t/s45 t/s53 t/s62 t/s
Fireworksfireworks/us74 t/s89 t/s103.2 t/s114 t/s
Fireworksfireworks/fast67 t/s93.5 t/s118.6 t/s147.72 t/s
Fireworksfireworks39 t/s43 t/s46 t/s53 t/s
InferenceNetinference-net/fast63 t/s88 t/s112 t/s163.44 t/s
InferenceNetinference-net/fp485 t/s114 t/s144 t/s211 t/s
Makoramakora/fp452 t/s67 t/s80.9 t/s96.79 t/s
Modalmodal/mxfp479 t/s103 t/s131 t/s223 t/s
Moonshot AImoonshotai/mxfp424 t/s32 t/s37 t/s44 t/s
Morphmorph/fp850 t/s64 t/s81 t/s113.6 t/s
Parasailparasail/fp465 t/s92 t/s126 t/s188 t/s
Parasailparasail/fast65 t/s86 t/s130 t/s162.02 t/s
Phalaphala49 t/s68 t/s93 t/s149 t/s
Relacerelace/fp452 t/s79 t/s114 t/s182.33 t/s
Sail Researchsail-research/fp437 t/s62 t/s85 t/s134 t/s
Togethertogether48 t/s56 t/s64 t/s81 t/s
Waferwafer47 t/s59 t/s71 t/s95 t/s
Waferwafer/us50 t/s63 t/s78 t/s114 t/s

Percentiles describe the observed request distribution. These are snapshots of a rolling window, not a historical time series.

Uptime

Compare successful request rates across the latest 5-minute, 30-minute, and 24-hour observations.

Successful requests by window
Availability by endpointCurrent overlapping windows · 25 endpoints
Endpoint5 minutes30 minutes24 hours
AkashMLakashml/fp4
Alibabaalibaba
Amazon Bedrockamazon-bedrock/us-east-2
BaseTenbaseten/fp8
Chuteschutes/mxfp4
Decartdecart/mxfp4
DeepInfradeepinfra/mxfp4
DigitalOceandigitalocean
Fireworksfireworks/us
Fireworksfireworks/fast
Fireworksfireworks
InferenceNetinference-net/fast
InferenceNetinference-net/fp4
Makoramakora/fp4
Modalmodal/mxfp4
Moonshot AImoonshotai/mxfp4
Morphmorph/fp8
Parasailparasail/fp4
Parasailparasail/fast
Phalaphala
Relacerelace/fp4
Sail Researchsail-research/fp4
Togethertogether
Waferwafer
Waferwafer/us

Reported by OpenRouter at the time of collection; rate-limited requests are excluded. A dash means no measurement was reported. This is not a live availability check.

Benchmarks

Review attributed evaluation results on each source's original scale.

External evaluations
moonshotai/kimi-k3Observed
Benchmark profileArtificial Analysis · shared source scale from 0 to 80
Coding76.2
Agentic50
Intelligence43.6
Design ArenaCategoryEloWin rateRank
agentsagenticgamedev1,25255%3
agentsandroidnative1,26255.4%5
agentsfullstack1,31561.8%2
agentsgodotgamedev1,22348.2%14
agentshtmlslides1,25659.6%2
agentsmobileapps1,26356.9%4
agentspython-pptxslides1,27359.2%4
agentswebapps1,28559.1%4
models3d1,39567.6%4
modelscodecategories1,36763%2
modelsdataviz1,35363.2%4
modelsgamedev1,36761%5
modelssvg1,30559.8%4
modelsuicomponent1,35561.3%4
modelswebsite1,34159.8%2

Sources: Artificial Analysis and Design Arena, via OpenRouter. Scores retain their original scales and are not OpenAI Suite ratings.

Detailed evaluations

Evaluation scoresAccuracy or item F1 · higher is better
gpqa diamondMoonshotAI: Kimi K3
91.928%
tau bench verified airlineMoonshotAI: Kimi K3
72.8%
EvaluationConfigurationSourceResultsDetails
gpqa diamondMoonshotAI: Kimi K3moonshotai/kimi-k3-20260715OpenRouter91.928%
Details
tau bench verified airlineMoonshotAI: Kimi K3moonshotai/kimi-k3-20260715OpenRouter72.8%
Details
Composite indicesKimi K3 (Max)moonshotai/kimi-k3-20260715Artificial AnalysisIntelligence 43.6 · Coding 76.2 · Agentic 50
Details
models · datavizKimi K3moonshotai/kimi-k3-20260715Design Arena1,351 Elo
Details
models · 3dKimi K3moonshotai/kimi-k3-20260715Design Arena1,389 Elo
Details
models · uicomponentKimi K3moonshotai/kimi-k3-20260715Design Arena1,340 Elo
Details
models · gamedevKimi K3moonshotai/kimi-k3-20260715Design Arena1,362 Elo
Details
models · codecategoriesKimi K3moonshotai/kimi-k3-20260715Design Arena1,360 Elo
Details
models · websiteKimi K3moonshotai/kimi-k3-20260715Design Arena1,335 Elo
Details
models · svgKimi K3moonshotai/kimi-k3-20260715Design Arena1,302 Elo
Details

Sources: Artificial Analysis, Design Arena and OpenRouter. Evaluation configurations and scales differ; results are not interchangeable. Updated Oct 24, 2026.

Apps & session costs

Compare observed median session costs for applications using this model, grouped by conversation length.

30-day sample · through Oct 18, 2026
ApplicationModel routeTurnsMedian cost / sessionDetails
Kilo Codemoonshotai/kimi-k3-2026071550+ turns
$4.779162
Details
Claude Codemoonshotai/kimi-k3-202607151 turn
$0.040191
Details
Hermes Agentmoonshotai/kimi-k3-202607152–9 turns
$0.080869
Details
Kilo Codemoonshotai/kimi-k3-202607152–9 turns
$0.191184
Details
Kilo Codemoonshotai/kimi-k3-2026071510–49 turns
$0.856235
Details
Claude Codemoonshotai/kimi-k3-2026071550+ turns
$4.944643
Details
Claude Codemoonshotai/kimi-k3-202607152–9 turns
$0.116768
Details
Claude Codemoonshotai/kimi-k3-2026071510–49 turns
$0.824566
Details
Hermes Agentmoonshotai/kimi-k3-2026071550+ turns
$3.422617
Details
Codexmoonshotai/kimi-k3-202607152–9 turns
$0.081447
Details
Hermes Agentmoonshotai/kimi-k3-2026071510–49 turns
$0.568668
Details
Codexmoonshotai/kimi-k3-202607151 turn
$0.013271
Details
Hermes Agentmoonshotai/kimi-k3-202607151 turn
$0.015979
Details

Source: OpenRouter, as of Oct 20, 2026. Published under CC BY 4.0. These are observed costs per session, not token prices or a forecast for your workload. Apps are compared separately.

Activity

Daily prompt and completion tokens reported for this model's routes in OpenRouter's top 50. Missing days have no published value.

Last 90 complete UTC days
Daily token usage20,961,518,627,079 reported tokens · 89 days with observations
Jul 26, 2026Oct 23, 2026
Daily observations · 89 days
Date (UTC)Reported tokensReported routes
Oct 23, 2026341,783,929,7221
Oct 22, 2026413,236,310,7211
Oct 21, 2026285,359,271,0541
Oct 20, 2026278,502,113,2301
Oct 19, 2026283,384,167,1841
Oct 18, 2026250,191,500,9361
Oct 17, 2026179,817,658,1441
Oct 16, 2026203,418,976,6041
Oct 15, 2026236,085,134,2961
Oct 14, 2026393,654,343,8061
Oct 13, 2026497,238,529,2601
Oct 12, 2026452,261,046,0081
Oct 11, 2026308,653,041,9971
Oct 10, 2026307,169,340,8831
Oct 9, 2026338,553,317,9711
Oct 8, 2026328,787,105,8851
Oct 7, 2026285,885,043,3081
Oct 6, 2026281,721,532,4791
Oct 5, 2026299,471,440,0101
Oct 4, 2026225,109,546,1281
Oct 3, 2026210,928,154,2871
Oct 2, 2026211,604,184,5501
Oct 1, 2026203,902,650,9731
Sep 30, 2026263,544,817,2991
Sep 29, 2026275,669,934,1811
Sep 28, 2026228,542,070,9561
Sep 27, 2026203,339,351,4801
Sep 26, 2026185,196,307,5941
Sep 25, 2026211,331,435,6301
Sep 24, 2026195,628,585,8261
Sep 23, 2026188,682,741,0831
Sep 22, 2026192,626,651,5571
Sep 21, 2026217,507,005,7401
Sep 20, 2026173,315,800,8651
Sep 19, 2026192,220,446,2931
Sep 18, 2026230,212,969,9901
Sep 17, 2026222,257,940,0941
Sep 16, 2026226,330,895,0591
Sep 15, 2026204,411,193,0191
Sep 14, 2026203,962,504,5731
Sep 13, 2026167,235,049,2611
Sep 12, 2026167,450,711,4461
Sep 11, 2026203,231,210,8681
Sep 10, 2026197,083,498,5901
Sep 9, 2026254,307,352,0041
Sep 8, 2026258,659,434,4881
Sep 7, 2026223,578,066,8491
Sep 6, 2026179,719,297,5971
Sep 5, 2026296,311,315,6391
Sep 4, 2026321,216,396,8621
Sep 3, 2026341,422,677,1441
Sep 2, 2026295,747,954,4591
Sep 1, 2026304,647,563,8101
Aug 31, 2026240,776,882,1411
Aug 30, 2026230,510,697,2881
Aug 29, 2026170,222,788,8171
Aug 28, 2026225,998,184,2011
Aug 27, 2026247,725,381,8611
Aug 26, 2026237,565,016,9331
Aug 25, 2026236,868,165,7101
Aug 24, 2026225,409,049,7761
Aug 23, 2026159,871,280,4451
Aug 22, 2026172,118,663,9221
Aug 21, 2026213,772,516,2801
Aug 20, 2026210,125,954,4591
Aug 19, 2026205,104,101,7791
Aug 18, 2026177,612,676,4161
Aug 17, 2026183,651,256,5821
Aug 16, 2026141,264,448,2461
Aug 15, 2026143,548,790,0601
Aug 14, 2026218,586,584,4101
Aug 13, 2026240,856,560,1411
Aug 12, 2026210,709,205,6481
Aug 11, 2026231,576,201,3671
Aug 10, 2026221,841,516,7681
Aug 9, 2026187,686,655,9881
Aug 8, 2026162,698,945,4971
Aug 7, 2026194,977,360,2101
Aug 6, 2026227,444,677,3621
Aug 5, 2026219,552,377,4131
Aug 4, 2026206,031,713,9831
Aug 3, 2026151,465,387,2931
Aug 2, 2026175,413,168,9471
Aug 1, 2026194,010,709,3191
Jul 31, 2026217,494,749,8331
Jul 30, 2026191,158,721,4591
Jul 29, 2026241,853,719,1701
Jul 28, 2026219,310,827,9291
Jul 27, 2026184,594,169,7341

Source: OpenRouter (openrouter.ai/rankings), as of Oct 24, 2026. CC BY 4.0. Totals include only individually published routes. Tokenizers differ by provider; the aggregated “other” category is never assigned to a model.

Technical details & API

Explore specifications, supported parameters, reasoning controls, and API identifiers for each OpenRouter route.

Current route specifications
moonshotai/kimi-k3Observed
Context
1,048,576
Maximum output
943,718
Tokenizer
Other
Input
text, image, video
Output
text
Moderation
No
Added to OpenRouter
2026-07-16
Base URLhttps://openrouter.ai/api/v1
Model IDmoonshotai/kimi-k3
Version IDmoonshotai/kimi-k3-20260715
Weights IDmoonshotai/Kimi-K3

Reasoning

Required
No
Enabled by default
Yes
Default effort
max
Supported efforts
max, high, low

Supported parameters

  • frequency_penalty
  • include_reasoning
  • logit_bias
  • logprobs
  • max_tokens
  • min_p
  • presence_penalty
  • reasoning
  • reasoning_effort
  • repetition_penalty
  • response_format
  • seed
  • stop
  • structured_outputs
  • temperature
  • tool_choice
  • tools
  • top_k
  • top_logprobs
  • top_p

Default parameters

top p
0.95

Frequently asked questions

Answers based on the model metadata and serving offers currently in the catalog.

How much does Kimi K3 cost?

Listed rates start at $0.33 per million input tokens and $9.90 per million output tokens. Prices vary by provider and configuration; see Pricing for details.

Which providers offer Kimi K3?

There are 133 cataloged offers across 82 providers. Compare model IDs, prices and limits in Providers.

What is the context window?

The cataloged model context is 1,048,576 tokens. Each serving endpoint may apply a different limit.

Does it support tools and structured output?

Tool calling: Not reported. Structured output: Not reported. Support can vary by endpoint.

Catalog history · 3 recorded revisions

Metadata observations since this model was first cataloged. These are not model release versions.

  • · a16d3344dbd8
  • · 0a125e654385
  • · c7865f02a448