What Moonshot AI does

Moonshot AI is a Beijing-based AI foundation model company developing the Kimi family of large language models and consumer/enterprise AI products. Founded in 2023 by Yang Zhilin, Xinyu Zhou, and Wu Yuxin, the company builds natively multimodal MoE models at the 1-trillion-parameter scale (K2.5, K2.6, K2.7), with proprietary technical innovations including the Attention Residuals architecture, Agent Swarm orchestration of up to 300 parallel sub-agents executing 4,000 coordinated steps, 90% KV cache compression, and the FlashKDA CUDA kernel. Its core products span the Kimi chatbot consumer application (iOS/Android, web, desktop), the Kimi Work desktop AI agent, the Kimi Code and K2.7 Code coding models, and an open platform API at platform.kimi.ai, with open-weight model releases under a Modified MIT License on Hugging Face.

The company's business model combines consumer subscription tiers (Kimi Work Moderato at $19/month and Vivace at $199/month), usage-based API pricing ($0.95 input / $4 output per million tokens — reported as 94-95% cheaper than US peers), enterprise AI services for large organizations, and open-source releases that drive developer adoption funnels into paid API and enterprise contracts. Go-to-market operates through four reinforcing motions: product-led growth via the Kimi chatbot freemium model, API-first distribution through OpenRouter and direct platform access, enterprise field sales targeting organizations like Dell, Cerebras, CoreWeave, ByteDance, and Tencent QQ Mail, and community-led adoption via open-source releases and developer relations. Customer segments span individual developers and knowledge workers (primary), enterprise customers integrating AI into workflows (primary), and AI developers building applications on Kimi models.

As of mid-2026, Moonshot AI had raised over $3.9 billion across funding rounds led by IDG Capital, Alibaba, Tencent, Meituan's Long-Z Investments, and China Mobile, with valuation rising from $2.5 billion in early 2024 to $30 billion pre-money by June 2026. Annual recurring revenue surpassed $200 million in April 2026, international revenue reached 62% of total income, and the company announced a red-chip corporate restructuring to prepare for a Hong Kong IPO. Distribution partnerships with Dell Enterprise Hub, Cerebras, CoreWeave, Huawei Cloud, and SiliconFlow, combined with named enterprise integrations (Cursor, ByteDance Doubao, Tencent QQ Mail Agently Mail), position the company across consumer, developer, and enterprise markets globally.

Name

Moonshot AI

Legal name

北京月之暗面科技有限公司

Website

https://moonshot.ai

Company type

Private

Founded year

2023

Operating status

Operating

Headcount range

251–500 employees

Short description

Moonshot AI is a Beijing-based AI foundation model company building the Kimi family of 1-trillion-parameter multimodal large language models and AI products. It serves developers, enterprises, and consumers through open-weight model releases, usage-based APIs, and consumer subscription tiers for its chatbot and desktop agent offerings.

Ownership category

akta.pro rank

Product category

Large Language Models / AI Foundation Models

NAICS

Software Publishers (5132), Software Publishers (513210)

SIC

Services-Prepackaged Software (7372)

akta.pro primary industry

Agents & Autonomous Workflows (Tool Use, Planning, Multi-Agent) (HDAAACAF)

akta.pro secondary industries

AI Application Enablement Platforms (Copilot/Agent Frameworks, SDKs) (HDAEANAJ), AI Integration & Orchestration Platforms (Connectors, Workflow, iPaaS for AI) (HDAEANAI), B2B/EDI & Partner Integration (BPAMAOAJ)

Keywords

• Large language models
• AI assistant platform
• Multimodal AI models
• Open-source AI
• Enterprise AI agents

Headquarters

HQ city

Haidian

HQ country

China

HQ region

Asia

Offices1 record

Markets served

GTM type

B2B and B2C

Offering type

Software

Cost components

Technology or R&D, Infrastructure, Personnel, Marketing or Sales, Operations

Revenue model

1. API Consumption: Revenue from developers and enterprises consuming AI model APIs through platform.kimi.ai. Usage-based pricing per million tokens. Major growth driver with fourfold increase in overseas API revenue following K2.5 release.
2. Kimi Chatbot Subscriptions: Consumer subscription tiers for Kimi chatbot (Moderato $19/month, Vivace $199/month for Kimi Work). Individual subscription payment orders surged 8,280% month-over-month in January 2026. Free tier drives adoption with conversion to paid.
3. Enterprise AI Services: B2B enterprise licensing and services including Kimi Work desktop application, API enterprise agreements, and model services. ARR topped $200 million in April 2026 driven by enterprise adoption.
4. Open-source Model Licensing: Open-weight models released under Modified MIT License. While free for research/use, generates ecosystem value and drives adoption that converts to paid API and enterprise services.

Pricing tiers

Go-to-market motion4 records

Distribution channels5 records

Marketing channels6 records

Core offering

Moonshot AI develops and sells the Kimi family of large language models and AI assistants. Its core products include the Kimi consumer chatbot (kimi.com), the Kimi Work desktop AI agent for enterprise productivity, and the platform.kimi.ai API giving developers programmatic access to Kimi K2.5/K2.6/K2.7 models. Open-weight model releases on Hugging Face drive ecosystem adoption that converts into paid API and enterprise contracts.

Product overview

Moonshot AI operates as a unified AI technology company centered on the Kimi AI assistant platform, supported by a portfolio of open-source and proprietary large language models. The core offering includes Kimi (consumer and enterprise AI assistant), with underlying model releases including Kimi K2.6 (current flagship with 1 trillion parameters, multimodal, and Agent Swarm supporting 300 parallel agents), Kimi K2.7 and K2.7 Code (coding-focused variants), and Kimi K2.5 (previous generation). Supporting products include Kimi Work (desktop agent with subscription tiers for up to 300 agents), Kimi Claw (OpenClaw-based agent framework), and Kimi Code (coding assistance). The Open Platform API enables developer access to Kimi models for enterprise and application development. All flagship models are released as open-source under Modified MIT License via Hugging Face.

Differentiator

Problem solved

Functional benefit

Brands

• Kimi: AI assistant chatbot product offering conversational AI, document analysis, and various productivity features
• Kimi API
• Kimi Work
• Kimi K2
• Kimi Claw
• Agent Swarm

Products and services

• Kimi Chatbot Consumer AI assistant offering conversational AI, document analysis, code assistance, and multimodal capabilities. Available via web (kimi.com), iOS/Android apps, Chrome extension, and desktop apps with freemium model plus Moderato ($19/month) and Vivace ($199/month) subscription tiers.
• Kimi K2.6 Open-source large language model with 1 trillion total parameters (32B activated per token) featuring native multimodal capabilities, 384-expert Mixture-of-Experts architecture, SwiGLU activation, and Agent Swarm support for up to 300 parallel agents. Scores 54 on HLE-Full doctorate-level test, outperforming GPT-5.4 and Claude Opus 4.6.
• Kimi Work Desktop AI agent application for macOS and Windows that runs locally with file access, browser control, and scheduled tasks. Features Agent Swarm capability spinning up to 300 sub-agents in parallel. Subscription tiers from $19/month (Moderato) to $199/month (Vivace).
• Kimi K2.7 Open-source AI model with 1 trillion parameters and 256k context window, positioned as a cost-effective alternative to leading models like Opus 4.8 and GPT-5.5 with multimodal capabilities.
• Kimi K2.7 Code Open-source AI coding model featuring a one-trillion-parameter Mixture-of-Experts architecture designed for complex programming tasks and agentic programming. Demonstrates strengths in dynamic web component generation and front-end design.
• Kimi K2.5 Open-source large language model that drove significant commercial growth, with revenue in the first 20 days post-release exceeding total 2025 revenue. Outperformed GPT-5 and Claude on benchmarks; boosted overseas API revenue by nearly fourfold. Used as base model for Cursor's Composer 2.
• Kimi Open Platform API Developer API platform at platform.kimi.ai enabling programmatic access to Kimi models for enterprise AI services and application development. Usage-based pricing at $0.95 per million input tokens and $4 per million output tokens.

Quantifiable outcome

• Revenue reached $120 million in first 20 days of 2026, surpassing full-year 2025 total of $115 million
• +4 more outcomes

Named customers6 records

Segments3 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration1 record

AI capability7 records

Feature8 records

Partnerships

Six partnerships are on record, tiered major and minor.

• Huawei CloudmajorStrategic or Co-development Partner · 5 June 2026Partnered with over 20 model developers including Moonshot AI to address vertical productivity use cases in healthcare, embodied intelligence, and confidential computing for government and financial sectors.
• CerebrasmajorTechnology or Integration · 20 May 2026Running Kimi K2.6 at 981 output tokens per second, 6.7x faster than next-fastest GPU-based cloud provider, positioning for enterprise inference market.
• Dell TechnologiesmajorTechnology or Integration · 18 May 2026Added Kimi K2.6 to Dell Enterprise Hub platform optimized for PowerEdge XE9780 servers with 8x NVIDIA HGX B300 GPUs, targeting enterprise on-premises AI workloads.
• Global Mofy AI LimitedminorStrategic or Co-development Partner · 13 May 2026Strategic investment in financing round as part of expansion into generative AI foundation models and AI ecosystems, enhancing AI-powered content generation capabilities.
• CoreWeavemajorTechnology or Integration · 11 May 2026Achieved #1 ranking for inference speed and price-performance for Kimi K2.6 in independent benchmark by Artificial Analysis, delivering 205 tokens/sec at $0.7 per million tokens.
• CursormajorTechnology or Integration · 23 March 2026Cursor built its Composer 2 coding model using Kimi K2.5 as base, accounting for approximately 25% of training volume. Moonshot publicly questioned Cursor's initial non-disclosure of base model provenance.

Scale indicators10 records

Recent moves8 records

Expansion highlights7 records

Direct peers

• DeepSeek: Chinese AI lab developing open-weight frontier LLMs (DeepSeek-V3, R1) with aggressive pricing. Most direct comparable to Moonshot: same open-weight strategy, Chinese base, competing on cost-performance against US frontier labs, and similarly named in US distillation concerns.
• Zhipu AI: Beijing-based AI company developing GLM series foundation models with consumer (ChatGLM) and enterprise (BigModel) offerings. Direct competitor to Moonshot in the Chinese frontier LLM market with overlapping customer base and similar state-backed investor profile.
• Alibaba (Qwen): Qwen model family from Alibaba Cloud is a leading open-weight frontier LLM. Directly competes with Kimi on benchmarks, pricing, and global developer mindshare; also a strategic investor in Moonshot, creating both partnership and competitive tension.
• OpenAI: Creator of GPT-5 series and ChatGPT, the global frontier model benchmark Moonshot explicitly targets. Both compete on enterprise API, consumer chatbot, and agent capabilities; OpenAI sets the price ceiling that Moonshot undercuts by 94-95%.
• Anthropic: Developer of Claude Opus 4.6, the model Kimi K2.6 reportedly outperforms on HLE-Full. Direct frontier competitor with overlapping enterprise positioning and coding agent focus (Claude Code); also named alongside Moonshot in Tencent QQ Mail integrations.
• ByteDance (Doubao): ByteDance's Doubao LLM family powers the consumer-facing Doubao chatbot and enterprise agent platform. Major Chinese frontier competitor with massive distribution; simultaneously a Moonshot customer via Kimi Work integration into Doubao Super Mode.
• 01.AI: Chinese AI lab founded by Kai-Fu Lee developing the Yi open-weight model family. Comparable as another well-funded Chinese open-weight LLM startup competing for the same enterprise and developer mindshare in domestic and overseas markets.
• xAI: Elon Musk's AI lab developing Grok frontier models with aggressive API pricing. Comparable to Moonshot as a recent entrant challenging US frontier incumbents with cost-disruption and rapid scaling.

Broad incumbents

• Meta (Llama): Meta's Llama open-weight models set the original open-weight frontier benchmark. Competes indirectly with Moonshot for enterprise self-hosted deployments and developer mindshare, though Meta is a broader consumer/internet incumbent rather than a pure AI lab.

Emerging players

• Mistral AI: Paris-based open-weight LLM lab with strong European enterprise traction. Comparable as an emerging open-weight frontier competitor with API and self-hosted offerings; smaller scale than Moonshot but similar open-weight philosophy and enterprise focus.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat6 records

Key risks6 records

Key highlights7 records

Customer concentration

• Xx.com/Kimi_Moonshot
• GitHubgithub.com/MoonshotAI
• Instagraminstagram.com/kimiai.official
• LinkedInlinkedin.com/company/kimi-ai-linkedin

Revenue estimate

Valuation estimate

Number of profiles

Profiles3 records

Funding overview

Funding rounds17 records

Investors19 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Moonshot AI