Tool profile

Kimi

ByMoonshot AICN

Moonshot AI assistant for research, documents, coding and long-running tasks, powered by multimodal, long-context models.

Kimi symbol in white on a dark gradient background
Country
CN
Availability
Proprietary
Modalities
  • Text
  • Image
  • Video
  • Code

Kimi is the AI assistant from Moonshot AI (Beijing, China), a company founded in 2023. Its flagship model, Kimi K3, has 2.8 trillion parameters, open weights, a 1-million-token context window and native understanding of images and videos, built for deep reasoning, long-horizon coding and knowledge work.

What it does

  • Conversations, writing and answers with advanced reasoning
  • Analysis of long documents (contracts, research, reports) thanks to the large context
  • Assisted coding and code tasks (Kimi Code)
  • Deep research and multi-task agents (Deep Research, Swarm)
  • Work with documents, spreadsheets, slide decks, websites and design (Docs, Sheets, Slides, Websites, Design)

How it works

  • You chat or describe a task
  • The model processes with multi-step reasoning and long context
  • Agent tools handle tasks such as research, file analysis and content generation
  • Paid subscriptions add agent credits, higher priority and parallel tasks

Models

The Kimi family is made by Moonshot AI. Kimi K3 has 2.8 trillion parameters, open weights, a 1-million-token context window and native vision (images and videos). There are also specialized models such as Kimi K2.7 Code (coding, 256k context) and earlier versions (K2.6, etc.). The platform API is compatible with OpenAI and Anthropic formats.

Public pricing

  • Free access is available (with usage limits)
  • Paid subscriptions with additional limits and features (agent credits, faster models, higher concurrency)
  • API is billed separately, pay-as-you-go
  • Plan names and prices may vary by region and billing cycle; check the official page

Strengths

  • Very long context (1 million tokens) for extensive documents and tasks
  • Large-scale open-weights model with strong reasoning
  • Integrated agent and tool ecosystem (Docs, research, coding, websites)
  • API compatible with common standards

Points of attention

  • Chinese company: review data, privacy and regulatory policies before professional use
  • Global availability of the app and plans varies; confirm on the official page
  • Features and prices change quickly (Moonshot has already retired older models)
  • Performance claims (“largest open model”, “first in its class”) come from the company and should be contextualized