Skip to content
  • Models
  • Rankings
  • Ori
Sign Up
Sign Up
OpenRouterOpenRouter
© 2026 OpenRouter, Inc

Product

  • Chat
  • Rankings
  • Benchmarks
  • Apps
  • Discover
  • Models
  • Collections
  • Providers
  • Pricing
  • Business
  • Enterprise
  • Labs

Company

  • About
  • Blog
  • Careers
    Hiring
  • Privacy
  • Terms of Service
  • Trust Center
  • Support
  • Works With OR
  • Data
  • Brand

Developer

  • Documentation
  • API Reference
  • Developer Platform
  • Status

Connect

  • Discord
  • GitHub
  • LinkedIn
  • X
  • YouTube
Favicon for moonshotai

MoonshotAI: Kimi K3

moonshotai/kimi-k3

Model weights
Compare

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.

Modalities

In / Out Price

50% off

$1.50 / $7.50per 1M

Context

1.0M

Released

Jul 16, 2026

Compare
BenchmarksProvidersPricingPerformanceUptimeAppsActivityFAQExplore

Benchmarks

Scores on standardized evaluations and this model’s own outputs on shared benchmark prompts. Higher percentages are better — and each rank shows where this model lands among the models evaluated on that source.

Benchmark score summary for MoonshotAI: Kimi K3 (Artificial Analysis and Design Arena)
SourceBenchmarkScore
Artificial AnalysisKimi K3 (low) Coding Index72.0
Artificial AnalysisKimi K3 (low) GPQA Diamond84.2%
Artificial AnalysisKimi K3 (low) HLE25.0%
Artificial AnalysisKimi K3 (low) AA-LCR79.3%
Artificial AnalysisKimi K3 (low) CritPt3.1%
Artificial AnalysisKimi K3 (low) SciCode52.7%
Artificial AnalysisKimi K3 (low) AA-Omniscience Accuracy45.8%
Artificial AnalysisKimi K3 (low) AA-Omniscience Non-Hallucination Rate22.9%
Artificial AnalysisKimi K3 (max) Intelligence Index43.6
Artificial AnalysisKimi K3 (max) Coding Index76.2
Artificial AnalysisKimi K3 (max) Agentic Index50.0
Artificial AnalysisKimi K3 (max) GPQA Diamond93.5%
Artificial AnalysisKimi K3 (max) HLE46.9%
Artificial AnalysisKimi K3 (max) AA-LCR88.7%
Artificial AnalysisKimi K3 (max) GDPval-AA51.2%
Artificial AnalysisKimi K3 (max) CritPt23.4%
Artificial AnalysisKimi K3 (max) SciCode59.5%
Artificial AnalysisKimi K3 (max) AA-Omniscience Accuracy47.6%
Artificial AnalysisKimi K3 (max) AA-Omniscience Non-Hallucination Rate46.8%
Design ArenaKimi K3 Agents Arena Agenticgamedev Elo1252
Design ArenaKimi K3 Agents Arena Androidnative Elo1259
Design ArenaKimi K3 Agents Arena Full Stack Elo1328
Design ArenaKimi K3 Agents Arena Godotgamedev Elo1199
Design ArenaKimi K3 Agents Arena Htmlslides Elo1259
Design ArenaKimi K3 Agents Arena Mobile Apps Elo1280
Design ArenaKimi K3 Agents Arena Python-Pptxslides Elo1273
Design ArenaKimi K3 Agents Arena Webapps Elo1296
Design ArenaKimi K3 Models Arena 3D Elo1423
Design ArenaKimi K3 Models Arena Code Categories Elo1384
Design ArenaKimi K3 Models Arena Data Visualization Elo1358
Design ArenaKimi K3 Models Arena Game Development Elo1407
Design ArenaKimi K3 Models Arena SVG Elo1329
Design ArenaKimi K3 Models Arena UI Component Elo1369
Design ArenaKimi K3 Models Arena Website Elo1351

Providers

Different companies host the same model. OpenRouter routes your request to one of them based on the routing mode you pick — Balanced (price + speed), Nitro (fastest), Floor (cheapest), or Exacto (highest tool-calling accuracy).

Pricing

The average price customers actually pay for this model, next to the prices providers post. Caching and discounts mean the price actually paid is often well below the listed one.

Performance

Throughput is how fast the model writes (tokens per second — higher is better). Latency is total round-trip time (lower is better). TTFT is time-to-first-token — how long before you see anything appear (lower is better).

Uptime

Uptime is the percentage of the past 3 days that at least one provider was responding to requests. Availability is the percentage of time that inference was successfully served. OpenRouter continuously monitors and uses the next-best provider when one returns an error.

Apps

Public apps that send the most traffic to this model. Good signal for what real production workloads look like — and a hint at which use cases this model is best suited for.

Activity

Token volume and request traffic to this model over time.

Quick Start

Drop-in code to call this model. OpenRouter's API is OpenAI-compatible — most SDKs work by just swapping the base URL. The only thing that changes between models is the model slug below.

Explore more models

AI Models with Vision: Multimodal LLMs for Image UnderstandingCollectionAI Model RankingsRanking
50% off
$3.00$1.500$15.00$7.500$0.30$0.15001.79s31 tps
100.00%
$1.700$8.500$0.17002.17s42 tps
89.58%
$2.100$10.950$0.21000.62s33 tps
100.00%
15% off
$2.50$2.125$14.00$11.900$0.29$0.24651.76s9 tps
100.00%
$2.144$10.758$0.24401.52s66 tps
100.00%
$2.850$14.250$0.28502.45s31 tps
99.42%
$3.000$15.000$0.30003.16s24 tps
96.15%
$3.000$15.000$0.30001.24s47 tps
99.61%
$3.000$15.000$0.30002.84s56 tps
100.00%
$3.000$15.000$0.30001.36s40 tps
100.00%
$3.000$15.000$0.30001.91s35 tps
100.00%
$3.000$15.000$0.30001.55s50 tps
93.81%
$3.000$15.000$0.30004.98s24 tps
100.00%
$3.300$16.500$0.33001.18s28 tps
100.00%
$3.450$17.250$0.34505.30s20 tps
96.00%
$4.500$22.500$0.45000.84s56 tps
100.00%
$6.000$22.500$0.60001.72s9 tps
100.00%
31% off
$3.00$2.070$15.00$10.350$0.30$0.20708.76s16 tps
82.91%
$2.550$12.750$0.25600.67s60 tps
100.00%
$2.550$12.950$0.25501.20s38 tps
99.15%

Throughput

66tok/s

P50, best across providers

Latency

0.62s

P50, best provider

AutoExacto Benchmarks
GPQA DiamondTAU-BenchFireworks Fast93.3%--Makora92.8%--Baseten92.2%--Chutes91.5%--Parasail93.6%73.5%Fireworks (US)93.1%71.1%Fireworks92.3%70.2%
+13 more providers
Uptime (3d)The model was reachable. Request routed to a provider.

100.00%

Availability (3d)The model returned inference from any provider. Errors and empty responses count against it.

99.65%

Availability over the last 3 days

Last 72 hours
Availability 99.65%
3 Days Ago2 Days AgoYesterdayNow

Availability over the last 24 hours

OpenRouter Availability
99.64%
Without Routing
96.51%

When an error occurs in an upstream provider, we can recover by routing to another healthy provider, if your request filters allow it. You can access per-provider uptime data programmatically through the Endpoints API. Learn more about our load balancing and customization options.

1.
Favicon for https://nousresearch.com
Hermes Agent
Hermes Agent is an open-source, self-improving AI agent by Nous Research that runs persistently with memory across sessions, and builds reusable skills from experience. It comes with 40+ built-in tools, including web search, browser automation, and vision, plus scheduled automations and subagents.
368Btokens
2.
Favicon for https://pi.dev/
pi
There are many coding agents, but this one is yours.
270Btokens
3.
Favicon for https://claude.ai/apple-touch-icon.png
Claude Code
Claude Code is Anthropic's agentic coding tool that reads your entire codebase, plans and executes changes across files, runs tests, and iterates on failures, all from natural language prompts.
259Btokens
4.
Favicon for https://afterquery.com/
ale-evol
Applied research lab for frontier models
159Btokens
5.
Favicon for https://docs.all-hands.dev/
OpenHands
AI coding agent that can run commands, browse the web, call APIs
103Btokens

Frequently asked questions

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.

Kimi K3 costs $1.50/M input tokens and $7.50/M output tokens, with separate rates for Cache Read at $0.15/M tokens.

Kimi K3 has a 1,048,576 token context window. It supports up to 1,048,576 completion tokens.

Yes. Kimi K3 accepts tools for function calling on 16 of the 17 providers serving it, and requests that send tools are routed to those providers. It also supports structured outputs via a JSON schema in response_format.

Kimi K3 accepts text, images, and video as input and returns text.

Kimi K3 is served by 17 providers on OpenRouter: inference.net, Relace, Phala, Wafer, Morph Fast, Sail Research, Makora, DigitalOcean and 9 more. Requests are routed to the best available provider, with automatic failover to the others, and you can pin or exclude providers with provider routing.

Kimi K3 was released on July 16, 2026.