LLM Pricing
ModelsLeaderboardsToolsProvidersReleasesArticles

© 2026 LLM Pricing

About
·Contact
·Privacy
·RSS
← Model list

Kimi K3

moonshotai·moonshotai/kimi-k3·BETA·Open weights·kimi-k3 series
Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work

Specs & pricing

Input / output per 1M tokens
Official price·Moonshot AI
$3.00 / $15.00
Blended $6.00 · Cache read $0.30
Lowest paid·UplinkSwitchGateway
$0.90 / $4.50
Blended $1.80 · 31.3× spread
2 more $0 channels
Context
1,048,576
Output limit
1,048,576
Knowledge cutoff
—
Released / updated
2026-07-16 / 2026-07-16
Capabilities
✓ Reasoning✓ Tool use✓ Structured outputTemperature✓ Attachments
⚠ Providers report capability flags inconsistently
Modalities
TextImageVideoPDF

Kimi K3 by moonshotai is offered by 131 providers on this page. Public prices are shown for 122 of them. Its official list price is $3.00 per 1M input tokens and $15.00 per 1M output tokens. The lowest paid channel is UplinkSwitch at $0.90 / $4.50 per 1M, about 3.3× below the list price. That channel is a third-party gateway, so confirm its availability and rate limits before depending on it. It also has 2 free ($0) channels and 6 covered by a paid subscription; free tiers usually carry rate limits, and subscription-covered access bills $0 per token only after the subscription fee.

Artificial Analysis rates it 17.4 on the Intelligence Index, with 37 for Strategy & Ops, its strongest domain. Against its lowest blended price of $1.80 per 1M, that is roughly 10 index points per dollar, which is the value ratio the leaderboards rank on. Median output speed is 44 tokens per second, with 4.15s to the first token. Running one task of the Artificial Analysis suite costs about $2.3688, which reflects how many tokens its reasoning consumes rather than the unit price alone.

The context window is 1,048,576 tokens at the reference host, but hosts report different limits, from 262,144 to 1,113,088, so the usable window depends on the provider you pick. It supports reasoning, tool use, and structured output. Accepted input modalities are Text, Image, Video, and PDF. The weights are open, so it can also be self-hosted or served through a gateway of your choice. Providers report the capability flags inconsistently, so verify a specific feature against the host you plan to use.

Quality & performanceArtificial Analysis · Intelligence Index v5 · rep. tier Low

Intelligence17.4
Value10
Engineering22.8
Finance & Accounting28.3
Legal27.7
Healthcare & Medical29.2
Strategy & Ops37
Economics27.7
Output speed44 tok/s
TTFT4.15 s
Cost per task$2.37
Value formulaIQ 17.4 ÷ min blended $1.800 = 10

Quality is independently evaluated by Artificial Analysis. Speed/latency are model-level medians.

Available at 131 providers122 with public prices · 2 free · 6 subscription-covered

  • ProviderInput / Output
  • Kimi For Coding (kimi.com)Gateway
    k3-256k
    Subscription
    Cache read $0Context 262,144 ⚠Output limit 262,144
  • Kimi For Coding (kimi.com)Gateway
    k3
    Subscription
    Cache read $0Context 1,048,576Output limit 1,048,576
  • NvidiaCloud
    Free
    Context 1,048,576Output limit 1,048,576deprecated
  • Kimi For Coding (kimi.ai)Gateway
    k3-256k
    Subscription
    Cache read $0Context 262,144 ⚠Output limit 262,144
  • Kimi For Coding (kimi.ai)Gateway
    k3
    Subscription
    Cache read $0Context 1,048,576Output limit 1,048,576
  • SCNet Token PlanGateway
    Kimi-K3
    Subscription
    Cache read $0Context 1,048,576Output limit 1,048,576
  • SenseNova (China)Gateway
    Free
    Cache read $0Context 1,048,576Output limit 65,536
  • Volcengine Ark Coding PlanGateway
    Subscription
    Cache read $0Context 1,048,576Output limit 1,048,576
  • UplinkSwitchGateway
    $0.90 / $4.50
    Cache read $0.12Context 1,048,576Output limit 1,048,576
  • NanoGPTGateway
    $0.45 / $9.60
    Cache read $0.13Context 1,048,576Output limit 943,718
  • Moonshot AI (China)OfficialFirst-party
    $2.99 / $14.94
    Cache read $0.30Context 1,048,576Output limit 1,048,576
ProviderTierInputOutputCache readCache writeContextOutput limitStatus
Kimi For Coding (kimi.com)
k3-256k
GatewaySubscription$0$0262,144 ⚠262,144
Kimi For Coding (kimi.com)
k3
GatewaySubscription$0$01,048,5761,048,576
NvidiaCloudFree——1,048,5761,048,576deprecated
Kimi For Coding (kimi.ai)
k3-256k
GatewaySubscription$0$0262,144 ⚠262,144
Kimi For Coding (kimi.ai)
k3
GatewaySubscription$0$01,048,5761,048,576
SCNet Token Plan
Kimi-K3
GatewaySubscription$0—1,048,5761,048,576
SenseNova (China)GatewayFree$0—1,048,57665,536
Volcengine Ark Coding PlanGatewaySubscription$0—1,048,5761,048,576
UplinkSwitchGateway$0.90$4.50$0.12—1,048,5761,048,576
NanoGPTGateway$0.45$9.60$0.13—1,048,576943,718
Moonshot AI (China)OfficialFirst-party$2.99$14.94$0.30$2.991,048,5761,048,576

Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.

Reasoning control

effort = noneeffort = loweffort = higheffort = maxToggle (on / off)effort = mediumeffort = minimaleffort = xhighbudget_tokens ≥ 1,024

Interleaved thinking (reasoning between tool calls) is declared by 66 of 131 providers.

10 / 131 providers expose no reasoning control (reasoning_options: []).

Experimental modes3 items

ModeProviderInputOutputCache readCache write
priorityFireworks AI$3.00$15.00$0.30—
priorityAbliteratedLLM$3.38$16.88$0.34—
priorityFireworks AI$3.75$18.75$0.38—

Your usage cost

Common scenarios
1UplinkSwitch$311.40
2NanoGPT$531.60
3KoRouter$595.08
4Inferris$615.60
5NanoGPT$645.20
6Chutes$645.20
32Moonshot AI (China) · Official$1,021.86

8 more channels offer $0 (Kimi For Coding (kimi.com), Kimi For Coding (kimi.com), Nvidia etc.); free tiers usually have rate limits and no SLA, excluded from ranking.

Switch to UplinkSwitch to save $710.46/mo (70%)
Note: this is a gateway; verify availability and rate limits yourself.

Price historyone sample accumulated per data sync

Input listOutput listMin blended
$15.00$02026-08-052026-11-15
2026-11-15Input list $3.00Output list $15.00Min blended $1.80

Benchmark13 items

NameConditionsScoreMetricSource
DeepSWEharness: Kimi Code · variant: max effort · v1.167.5resolve rateSource ↗
Terminal-Benchharness: Kimi Code · variant: max effort · v2.188.3accuracySource ↗
FrontierSWEharness: Kimi Code · variant: max effort81.2dominance scoreSource ↗
Program Benchharness: Kimi Code · variant: max effort77.8scoreSource ↗
SWE Marathonharness: Claude Code · variant: max effort · v1.142resolve rateSource ↗
GDPval-AAvariant: max effort · vv21668EloSource ↗
AA-Briefcasevariant: max effort1548EloSource ↗
AutomationBenchvariant: max effort · dataset: 600-task public subset30.8success rateSource ↗
JobBenchvariant: max effort52.9scoreSource ↗
SpreadsheetBenchharness: Claude Code · variant: max effort · v234.8scoreSource ↗
BrowseCompvariant: max effort, context compaction91.2accuracySource ↗
CharXiv Reasoningvariant: max effort, with tools91.3accuracySource ↗
ZeroBenchvariant: max effort, with tools41pass@5Source ↗

The same benchmark scores very differently across harness / dataset, so the qualifying conditions must be shown together.

Related models

Kimi K3.1 Pro Abliterated (Lyber)same series—Kimi K3.1 Pro Highspeedsame series—Kimi K3.1 Prosame series$2.00 / $10.00GLM-5.3cheaper alternative$1.40 / $4.40GLM-5.3-Flashcheaper alternative$0.15 / $0.50DeepSeek V4.1 Flash 1017cheaper alternative$0.15 / $0.60GLM-5.2cheaper alternative$1.40 / $4.40Kimi K2.8 Previewsame lab—
Data partly from models.dev (MIT) · Moonshot AI official docs ↗