Kimi K3

Moonshot AIMCP Tool CallReasoning

moonshotai/kimi-k3Open Weights

Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work

Benchmark Pricing$3.00 / 1M in$15.00 / 1M out$0.30 cache read

Context: 1,048,576 tokens•Max Output: 131,072 tokens•Released: 2026-07-16

TECHNICAL SPECIFICATIONS

Context Window1,048,5761M tokens

Max Output131,072131.1K tokens

LicenseOpen SourceSelf-hostable

Input Modalities

TextVisionVideo

Function / Tool CallingSupported (MCP Ready)

Structured OutputsGuaranteed Schema

MODEL CONTEXT PROTOCOL (MCP) READINESS

✓ Kimi K3 natively supports OpenAI / Anthropic format tool calling and is fully compatible with MCP servers and agent loops.

MCP Context Window Recommendations

With a context capacity of 1,048,576 tokens, this model can handle extremely large MCP resource manifests, complete database schemas, and multiple active server tool definitions simultaneously without degradation.

STANDARDIZED BENCHMARKS & EVALS

DeepSWEresolve rate

67.5

Terminal-Benchaccuracy

88.3

FrontierSWEdominance score

81.2

Program Benchscore

77.8

SWE Marathonresolve rate

GDPval-AAElo

1668

AA-BriefcaseElo

1548

AutomationBenchsuccess rate

30.8

JobBenchscore

52.9

SpreadsheetBenchscore

34.8

BrowseCompaccuracy

91.2

CharXiv Reasoningaccuracy

91.3

ZeroBenchpass@5

API CALL INVOCATION EXAMPLE

// Example: Calling Kimi K3 via OpenAI SDK
import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.MOONSHOTAI_API_KEY,
});

const response = await client.chat.completions.create({
  model: "moonshotai/kimi-k3",
  messages: [{ role: "user", content: "Hello! Can you inspect MCP tools?" }],
});

console.log(response.choices[0].message.content);