Unlock exclusive introductory pricing for the newly launched Gemini 3.5 Flash.
M

Kimi K3

Indtast:$2.4/M
Output:$12/M
Kontekst:1,000k tokens
Udgivet:Jul 16, 2026

Kimi K3 is Kimi's flagship model, designed for long-range programming and end-to-end knowledge work, featuring 1M token context and leading-edge comprehensive intelligence.

Ny
Kommersiel brug

Playground for Kimi K3

Udforsk Kimi K3's Playground โ€” et interaktivt miljรธ til at teste modeller og kรธre forespรธrgsler i realtid. Prรธv prompts, juster parametre og iterer รธjeblikkeligt for at fremskynde udvikling og validere use cases.

Technical Specifications of Kimi K3

SpecificationKimi K3
DeveloperMoonshot AI
Model IDkimi-k3
Model typeFrontier Mixture-of-Experts LLM
Total parameters2.8 trillion
Experts896 experts, 16 activated per token
Context windowUp to 1,000,000 tokens
Native multimodalityVisual understanding; Kimi API docs include image and video input examples
Core architectureKimi Delta Attention, Attention Residuals, Stable LatentMoE
Main use casesCoding, software engineering, long-context reasoning, agent tasks, knowledge work, visual reasoning
Hosted API accessKimi API Platform, CometAPI
Open weightsScheduled by July 27, 2026

What is Kimi K3?

Kimi K3 is Moonshot AI's flagship large language model and currently the company's most capable foundation model. It is designed for professional software engineering, long-context reasoning, autonomous agents, and complex knowledge-intensive tasks. Unlike previous Kimi models, K3 expands the context window to 1 million tokens, enabling repository-scale coding, long document analysis, and multi-step agent workflows.

The model is built on a large-scale Mixture-of-Experts architecture and represents Moonshot's latest frontier model competing with leading proprietary systems from OpenAI and Anthropic.

Main Features of Kimi K3

  • 1M-token context window enables reasoning across massive codebases, books, research papers, and enterprise documentation without aggressive chunking.
  • Optimized for software engineering with strong performance on repository-level coding, debugging, UI implementation, and agentic development workflows.
  • Designed for AI agents capable of long-horizon planning and multi-step task execution instead of isolated prompt completion.
  • Reasoning-aware inference supports configurable reasoning effort for complex problem solving.
  • K3 introduces two major architectural innovations: Kimi Delta Attention (KDA) for efficient long-sequence processing; Attention Residuals (AttnRes) to improve information retrieval across model depth
  • Kimi K3 approximately 2.5ร— higher scaling efficiency compared with Kimi K2.
  • Open-weight release strategy, allowing broader research and deployment compared with closed proprietary frontier models.

Benchmark Performance of Kimi K3

Moonshot released one of the most comprehensive benchmark reports among recent frontier models. Public results indicate Kimi K3 is competitive with leading proprietary systems across software engineering, reasoning, and agent tasks.

BenchmarkKimi K3GPT-5.6 SolClaude Fable 5Claude Opus 4.8
TerminalBench 2.188.388.884.684.6
FrontierSWE81.271.386.666.7
ProgramBench77.877.676.871.9
SWE Marathon42.039.035.040.0
BrowseComp91.290.488.084.3
DeepSearchQA95.0โ€”94.293.1

Key observations

  • Kimi K3 exceeds GPT-5.6 Sol on FrontierSWE and BrowseComp.
  • It ranks among the strongest publicly reported coding models.
  • Long-context software engineering is one of its biggest strengths.
  • Agent benchmarks place it close to leading proprietary frontier systems, though Moonshot notes the strongest closed models still maintain an overall experience advantage.

Kimi K3

Limitations

  • Low- and high-reasoning modes are planned but were not available at launch.
  • Switching into K3 midway through an existing conversation may reduce answer quality because the model relies on complete reasoning history.
  • The model can occasionally be overly proactive when user intent is ambiguous.
  • Despite its strong benchmark results, Moonshot explicitly states that K3 does not surpass every proprietary frontier model across all tasks.

Why Use Kimi K3 on CometAPI?

For developers building with multiple foundation models, accessing Kimi K3 through CometAPI offers several operational advantages:

  • Unified API endpoint for Moonshot, OpenAI, Anthropic, Google, DeepSeek, and other providers.
  • Faster model switching without rewriting application logic.
  • Consistent authentication and request formats across providers.
  • Simplified billing and centralized usage management.
  • Easier evaluation of Kimi K3 against other frontier models within the same infrastructure.

Kimi K3 is especially valuable for organizations that require long-context reasoning, enterprise document intelligence, and advanced coding agents while retaining the flexibility of an open-weight model.

High-Value Kimi K3 API Use Cases on CometAPI

Coding agents and repository-scale debugging

Kimi K3's strongest public positioning is long-horizon coding. A CometAPI-powered coding agent can send the model repository maps, failing test output, error logs, relevant files, screenshots, and tool results. K3 can then propose a plan, call tools, inspect feedback, and iterate. Use streaming for interactive progress and preserve complete assistant messages between tool turns.

Enterprise document intelligence

The 1M context window is attractive for contracts, policies, filings, support logs, and technical manuals. A practical pattern is to use retrieval for everyday questions, then escalate to K3 when the user asks for cross-document synthesis, contradiction detection, audit prep, or a board-ready memo with citations.

Multimodal product QA

K3's visual understanding makes it useful for reviewing dashboards, landing pages, chart exports, design comps, bug screenshots, and rendered PDFs. Ask for specific outputs: accessibility issues, missing states, likely data-quality problems, mismatched copy, or prioritized bug reports.

Tool-using internal assistants

For internal operations, K3 can combine reasoning with tools such as CRM lookup, billing systems, documentation search, ticket retrieval, deployment status, and analytics queries. Use tool_choice to force retrieval when facts must come from systems of record, then switch back to auto once the model has the right tool context.

FAQ

Priser for Kimi K3

Udforsk konkurrencedygtige priser for Kimi K3, designet til at passe til forskellige budgetter og brugsbehov. Vores fleksible planer sikrer, at du kun betaler for det, du bruger, hvilket gรธr det nemt at skalere, efterhรฅnden som dine krav vokser. Opdag hvordan Kimi K3 kan forbedre dine projekter, mens omkostningerne holdes hรฅndterbare.

ModelComet Price (USD / M Tokens)Official Price (USD / M Tokens)Discount
kimi-k3
Indtast:$2.4/M
Output:$12/M
Indtast:$3/M
Output:$15/M
-20%

Eksempelkode og API til Kimi K3

Fรฅ adgang til omfattende eksempelkode og API-ressourcer for Kimi K3 for at strรธmline din integrationsproces. Vores detaljerede dokumentation giver trin-for-trin vejledning, der hjรฆlper dig med at udnytte det fulde potentiale af Kimi K3 i dine projekter.

#!/bin/bash

curl --retry 0 "https://api.cometapi.com/v1/chat/completions" \
  -H "Authorization: Bearer $COMETAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kimi-k3",
    "messages": [
      {
        "role": "system",
        "content": "You are Kimi, an AI assistant provided by Moonshot AI. You are proficient in Chinese and English conversations. You provide users with safe, helpful, and accurate answers. You will reject any questions involving terrorism, racism, or explicit content. Moonshot AI is a proper noun and should not be translated."
      },
      {
        "role": "user",
        "content": "Hello, my name is Li Lei. What is 1+1?"
      }
    ],
    "max_completion_tokens": 1024
  }'

cURL Code Example

#!/bin/bash

curl --retry 0 "https://api.cometapi.com/v1/chat/completions" \
  -H "Authorization: Bearer $COMETAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kimi-k3",
    "messages": [
      {
        "role": "system",
        "content": "You are Kimi, an AI assistant provided by Moonshot AI. You are proficient in Chinese and English conversations. You provide users with safe, helpful, and accurate answers. You will reject any questions involving terrorism, racism, or explicit content. Moonshot AI is a proper noun and should not be translated."
      },
      {
        "role": "user",
        "content": "Hello, my name is Li Lei. What is 1+1?"
      }
    ],
    "max_completion_tokens": 1024
  }'

Python Code Example

import os

from openai import OpenAI


client = OpenAI(
    api_key=os.environ["COMETAPI_KEY"],
    base_url="https://api.cometapi.com/v1",
    max_retries=0,
)

completion = client.chat.completions.create(
    model="kimi-k3",
    messages=[
        {
            "role": "system",
            "content": (
                "You are Kimi, an AI assistant provided by Moonshot AI. "
                "You are proficient in Chinese and English conversations. "
                "You provide users with safe, helpful, and accurate answers. "
                "You will reject any questions involving terrorism, racism, "
                "or explicit content. Moonshot AI is a proper noun and should "
                "not be translated."
            ),
        },
        {
            "role": "user",
            "content": "Hello, my name is Li Lei. What is 1+1?",
        },
    ],
    max_completion_tokens=1024,
)

print(f"Completion ID: {completion.id}")
print(completion.choices[0].message.content)

JavaScript Code Example

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.COMETAPI_KEY,
  baseURL: "https://api.cometapi.com/v1",
  maxRetries: 0,
});

const completion = await client.chat.completions.create({
  model: "kimi-k3",
  messages: [
    {
      role: "system",
      content:
        "You are Kimi, an AI assistant provided by Moonshot AI. " +
        "You are proficient in Chinese and English conversations. " +
        "You provide users with safe, helpful, and accurate answers. " +
        "You will reject any questions involving terrorism, racism, or explicit content. " +
        "Moonshot AI is a proper noun and should not be translated.",
    },
    { role: "user", content: "Hello, my name is Li Lei. What is 1+1?" },
  ],
  max_completion_tokens: 1024,
});

console.log(`Completion ID: ${completion.id}`);
console.log(completion.choices[0].message.content);

Uptime

Anmodningssuccessrate over de seneste 30 dage, der afspejler pรฅlideligheden hos hver modeludbyder. CometAPI overvรฅger alle tilsluttede udbydere i realtid, 24/7.

RespondLIVE
13171msAvg. Response
UptimeLIVE
100.0%Avg. Uptime

Versioner af Kimi K3

ร…rsagen til, at Kimi K3 har flere รธjebliksbilleder kan omfatte potentielle faktorer sรฅsom variationer i output efter opdateringer, der krรฆver รฆldre รธjebliksbilleder for konsistens, at give udviklere en overgangsperiode til tilpasning og migration, og at forskellige รธjebliksbilleder svarer til globale eller regionale slutpunkter for at optimere brugeroplevelsen. For detaljerede forskelle mellem versioner, henvises der til den officielle dokumentation.

Version
kimi-k3