MiniMax Agent

Chat

MiniMax

Product overview

Name of Agent: MiniMax Agent
Short description of agent: "MiniMax Agent is a general-purpose agent capable of completing complex, long-horizon tasks. Supported by a multi-agent system, it can formulate expert-level solutions through multi-step planning, flexibly break down task requirements, and execute multiple sub-tasks to deliver the final result." (link, archived)
Date of release: 09/06/2025: MiniMax Agent platform release (link, archived)
Advertised use: Code, PPT, Deep Research, Multimodal, MCP Ecosystem Integration (link, archived)
Monetisation/Usage price: Credit‑based subscription free 19 USD/month, basic 69 USD/month, pro
Who is using it?: end users, team plan
Category: Chat

Company & accountability

Developer: MiniMax
Name of legal entity: Shanghai Xiyu Jizhi Technology Co., Ltd. (link, archived)
Place of legal incorporation: Shanghai, China (link, archived)
For profit company?: Yes (link, archived)
Parent company?: Not applicable
Governance documents analysis: ToS (link, archived), Privacy Policy (link, archived)
AI safety/trust framework: None found
Compliance with existing standards: None found

Technical capabilities & system architecture

Model specifications: Powered by MiniMax M2 (link, archived)
Documention: (User Guide, archived)
Observation space: User input, file upload, internet access, MCP tools, multimodal input
Action space: Pre-built MCPs (e.g. Google Maps, GitHub/GitLab, Slack, Figma), Custom MCPs, multimedia generation, command line interface on virtual machines (link, archived)
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 - Users can browse all files generated from the task. Generated websites will be deployed. "All files used for the task, including research documents, code files, and summary reports, which are available for download. If a web page was deployed, a link to the deployed page will be provided for you as well." (link, archived)
User roles: Operator (directing the agent to complete tasks)
Component accessibility: MiniMax M2 Model is open-sourced Agent platform is close-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: The turn-based interaction paradigm by default requires user approval (issuing further instructions) to continue the interaction. Model can also ask follow-up/clarifying questions
Execution monitoring, traces, and transparency: Visible CoT and action trace documenting all activity
Emergency stop and shut down mechanisms and user control: User can pause/stop the agent at any time
Usage monitoring and statistics and patterns: Users can monitor available credits in account settings

Ecosystem interaction

Identify to humans?: None
Identifies technically?: None found
Interoperability standards and integrations: - MCP support (link, archived)
Web conduct: None found

Safety, evaluation & impact

Technical guardrails and safety measures: None found
Sandboxing and containment approaches: Server‑side execution: tasks run on MiniMax’s infrastructure; details on sandboxing or isolation are not disclosed.
What types of risks were evaluated?: None found
(Internal) safety evaluations and results: None found
Third-party testing, audits, and red-teaming: None found
Benchmark performance and demonstrated capabilities: Programming, tool use, and deep search benchmarks (SWE-Bench Verified, Multi-SWE-Bench, Terminal-Bench, ArtifactsBench, tau^2-Bench, GAIA text only, BrowseComp, FinSearchComp-global) (link, archived)
Bug bounty programmes and vulnerability disclosure: None found
Any known incidents?: None found
← Back to 2025 Index