Kimi OK Computer
ChatMoonshot AI
Product overview
Name of Agent: Kimi OK Computer
Advertised use: "Kimi is capable of working continuously for tens of minutes, it autonomously completes processes step-by-step—from requirements research, planning, and analysis to design and development. Kimi can directly deliver outputs such as feature-rich multi-page websites, mobile-friendly web applications, and high-quality editable presentations. Additionally, it can transform hundreds of thousands of rows of Excel data into visual, interactive displays, enabling you to derive insights from your data more easily." (link, archived) (translated by Gemini 3 Pro)
Monetisation/Usage price: Freemium with trial limits: during beta, users receive three trial attempts for free
Andante plan (OK Computer 10 uses) – RMB 49 per month
Moderato plan (OK Computer 20 uses each) – RMB 99 per month
Allegretto plan (OK Computer 40 uses each) – RMB 199 per month (link, archived)
Who is using it?: end users
Website: (https://www.kimi.com/ok-computer, archived)
Category: Chat
Company & accountability
Developer: Moonshot AI
Parent company?: Not applicable
AI safety/trust framework: None found
Compliance with existing standards: None found
Technical capabilities & system architecture
Model specifications: "Leveraging end-to-end reinforcement learning technology, we have trained the Kimi K2 model into a native Agent. This enables the model to master the proficient use of over 20 tools—including the file system, browser, terminal, coding, image generation, and audio generation—and to autonomously construct the intrinsic knowledge of "how to complete a complex task." (translated by Gemini 3 Pro; (link, archived))
Documention: No specific doc found
Observation space: User instructions, interaction history
Memory architecture: - Long‑term memory is used to retain history across interactions.
- No evidence of episodic memory across different conversations
User interface and interaction design: Chatbot, able to access the agent's planned TO DOs, intermediate files
User roles: Operator (directing the agent to complete tasks)
Examiner (user can use thumbs up/down buttons to give feedback)
Component accessibility: Closed source (trained based on Kimi K2, which is open-sourced)
Autonomy & control
Autonomy level and planning depth: L3: User as a consultant, “agent takes initiative in task planning and execution over extended time horizons”, user mainly gives feedback or just incorporates results.
User approval requirements for different decision types: User can offer text-based feedback in response to agent's questions
Execution monitoring, traces, and transparency: Highly abbreviated action trace (such as listing out all searched webpage, read to do, write to do, create markdown files), no visible CoT
Emergency stop and shut down mechanisms and user control: User can pause/stop the agent at any time
Usage monitoring and statistics and patterns: Unable to monitor usage statistics
Ecosystem interaction
Identify to humans?: None found
Identifies technically?: None found
Interoperability standards and integrations: None found
Web conduct: None found
Safety, evaluation & impact
Technical guardrails and safety measures: None found
Sandboxing and containment approaches: None found
What types of risks were evaluated?: None found
(Internal) safety evaluations and results: None found
Third-party testing, audits, and red-teaming: None found
Bug bounty programmes and vulnerability disclosure: None found
Any known incidents?: None found