Kimi

Kimi is an AI assistant and model platform from Moonshot AI that provides multimodal chat, deep reasoning, coding, research, agentic workflows, and developer APIs for working with complex tasks and large amounts of information.

Pricing: Free

Kimi AIAI AssistantMultimodal AIAI AgentsAI CodingDeep ResearchAI ReasoningLLM API

Kimi is an AI platform developed by Moonshot AI for individual users, developers, researchers, and businesses that need conversational AI, reasoning, coding, visual understanding, and agentic capabilities. The platform combines the Kimi web and mobile experience with an API platform that allows developers to integrate Kimi models into their own applications and workflows. The latest model family includes Kimi K2.6, a native multimodal model supporting text, image, and video inputs, thinking and non-thinking modes, conversations, coding, and agent tasks. It supports a 256K-token context window and features such as tool calling, JSON Mode, context caching, and internet search.

Kimi's consumer interface offers several modes, including K2.5 Instant, K2.5 Thinking, K2.5 Agent, and K2.5 Agent Swarm. The Agent Swarm capability can dynamically coordinate up to 100 sub-agents and as many as 1,500 tool calls for complex tasks, while Kimi K2.5 is designed for coding, visual reasoning, research, and office productivity. It can work with documents, spreadsheets, PDFs, presentations, images, and videos.

For developers, the Kimi API provides access to its model family through a Chat Completions interface and supports agent-oriented applications, tool calling, multimodal inputs, and official tools. Available official tools include web search, memory, Excel and CSV analysis, Python code execution, JavaScript execution, URL fetching, date processing, and unit conversion.

Kimi uses usage-based API pricing calculated primarily by tokens. The current K2.6 API price is $0.95 per million input tokens and $4 per million output tokens, with cached input priced at $0.16 per million tokens. Batch API inference is offered at 60% of standard model pricing.