AI assistants for documents, meetings, search and daily workflows.
DeepSeek
DeepSeek's English docs cover agent-tool integration as a documented path.
Best fit · Developers who want DeepSeek as a backend for coding assistants and terminal agents.
Coverage · 100/100
Qoder
QoderWake has its own installer, local console, role and memory model, automation API and multi-Waker groups, so it is a distinct agent runtime rather than another Qoder IDE feature.
Best fit · Individuals and teams testing persistent role-based AI workers for software delivery, product work, data analysis, content operations and event-driven automation.
Coverage · 100/100 · backfill: pricing
Alibaba Cloud
Token Plan is a distinct commercial route for Qwen Cloud and affects how overseas developers actually consume the models.
Best fit · Developers who want a subscription-style Qwen Cloud route for coding tools and compatible agents.
Coverage · 100/100 · backfill: sources
Moonshot AI
Kimi has shifted from a long-context assistant profile into a broader AI workspace and agent ecosystem, so the site should track both user-facing features and the Kimi API platform.
Best fit · Users and teams evaluating a Chinese AI assistant for long documents, deep research, presentations, sites, spreadsheets and agentic work.
Coverage · 100/100
Moonshot AI
Kimi Claw changes the product from chat to action, so it needs a separate risk and use-case profile.
Best fit · Teams researching Kimi's agentic execution and browser/workflow automation capabilities.
Coverage · 100/100
Moonshot AI
Kimi WebBridge is a distinct local-browser bridge in the Kimi agent ecosystem, so it deserves its own evaluation profile separate from Kimi chat, Kimi Code and Kimi Claw.
Best fit · Teams evaluating local browser automation for AI agents that need existing login sessions without moving page content to a cloud browser.
Coverage · 100/100
Moonshot AI
Kimi's visible product surface now includes concrete artifact workflows, which changes how overseas users should evaluate it.
Best fit · Users evaluating Kimi as an office-style AI workspace rather than only a chat assistant.
Coverage · 96/100 · backfill: sources
Zhipu AI
It is the English-facing Z.ai consumer entry listed on the official site.
Best fit · Users who want to try GLM through an English web assistant before using APIs.
Coverage · 96/100 · backfill: sources
Zhipu AI
ChatGLM remains a major Z.ai product name and should be separated from the developer API platform.
Best fit · Users evaluating Z.ai's consumer assistant experience, especially study and general AI chat.
Coverage · 100/100
Zhipu AI
AutoGLM is a distinct Z.ai agent line and should not be hidden inside the generic GLM API entry.
Best fit · Teams researching China-origin autonomous agents and app-operation workflows.
Coverage · 96/100 · backfill: sources
Zhipu AI
AutoClaw is now prominent on the English site and adds a local-agent product surface to the Z.ai map.
Best fit · Users testing a local Z.ai agent workspace with skills and browser control.
Coverage · 96/100 · backfill: sources
Zhipu AI
Zread.ai is a distinct developer-product entry on the English site, separate from GLM API and ChatGLM.
Best fit · Developers evaluating Z.ai for repository understanding and code discovery.
Coverage · 96/100 · backfill: sources
Zhipu AI
AMiner is an official product-family entry and helps distinguish research discovery from GLM assistant/API products.
Best fit · Users who need academic search and researcher discovery in the Z.ai ecosystem.
Coverage · 92/100 · backfill: sources, access signals
Zhipu AI
Zlearn is listed by the official English site and should not be omitted, but confidence is lower than BigModel or ChatGLM.
Best fit · Users tracking Z.ai's education product coverage at a high level.
Coverage · 88/100 · backfill: sources, pricing
Zhipu AI
Autotyper appears in the official English product list, but lacks enough public detail for a strong rating.
Best fit · Teams cataloging Z.ai productivity products before hands-on testing.
Coverage · 88/100 · backfill: sources, pricing
MiniMax
Mavis is a productized multi-agent runtime with background execution, verification and recoverable state, rather than only a general chat assistant.
Best fit · Teams delegating long-running coding, research and document workflows that benefit from parallel roles and independent verification.
Coverage · 100/100
MiniMax
Talkie belongs in the MiniMax international product map because the official site lists it as a product, but it is a consumer app rather than an API surface.
Best fit · Users comparing MiniMax's consumer character and companion AI distribution.
Coverage · 100/100
ByteDance / Volcano Engine
ArkClaw now has its own BytePlus product, pricing and positioning page, so it should be tracked separately from ModelArk Coding Plan and generic Doubao/Ark APIs.
Best fit · Teams and power users evaluating always-on AI colleagues for reporting, KPI monitoring, content operations, Lark workflows and ModelArk-backed automation.
Coverage · 100/100
AgentScope AI
AgentTeams belongs in Productivity because its core value is collaborative agent orchestration, not model hosting or a narrow coding workflow.
Best fit · Teams that need a self-hosted, auditable multi-agent operating layer for enterprise collaboration and task orchestration.
Coverage · 100/100
AgentScope AI
QwenPaw belongs in Productivity because it is a worker runtime for collaborative agent orchestration, not a direct model API or a full coding IDE.
Best fit · Teams that want a browser-friendly worker runtime inside a transparent multi-agent room rather than a monolithic assistant.
Coverage · 100/100 · backfill: sources
ByteDance / Volcano Engine
DeerFlow belongs in Productivity because it is a SuperAgent harness broader than an AI coding agent: its core promise is long-horizon research, creation and workflow execution through skills, sub-agents and sandboxes.
Best fit · Teams that want to self-host a Chinese-origin open-source agent harness for deep research, report generation, coding, file-based workflows and long-running multi-agent tasks.
Coverage · 100/100
Skywork AI
Skywork should be tracked as a workspace platform because the public surfaces are organized around task-specific agents and output formats, not one generic chat flow.
Best fit · Knowledge workers who want one cloud workspace for research, writing, slides, sheets, websites and short-form media output.
Coverage · 100/100 · backfill: access signals
Tencent Cloud
WorkBuddy belongs in Productivity because it is an office agent workspace rather than a general chat product or coding tool.
Best fit · Teams that want an office workspace agent for cross-app coordination, task execution and model routing.
Coverage · 100/100 · backfill: access signals
Tencent Cloud
Miora belongs in Productivity because it is an agentic workspace for creative production, not a standalone model API or a narrow media generator.
Best fit · Teams evaluating a multi-agent creative workspace for visual asset generation, video ideation, 3D concepts and prototype drafting.
Coverage · 100/100 · backfill: pricing, access signals
Microsoft
Copilot Cowork is a useful benchmark for comparing Chinese agent subscriptions because it makes enterprise autonomous-work usage explicitly metered.
Best fit · Microsoft 365 enterprises that want governed long-running agents across mail, calendar, files, Teams, Dynamics and connected tools.
Coverage · 100/100
ByteDance / Volcano Engine
Dola expands the site beyond developer APIs by covering China-linked international consumer assistant distribution.
Best fit · Overseas consumer users who want a free Chinese-built AI assistant with low registration friction.
Coverage · 100/100
Lumina Lab
Tabbit matters for this directory because it exposes Chinese and global AI models inside an agentic browser workflow, with context and Skills as the product boundary.
Best fit · Knowledge workers, researchers and creators who want browser-native AI agents with tab, file and saved-page context.
Coverage · 100/100