OpenClaw was built to be model-agnostic. You can swap the LLM backend without changing your skills, your messaging integrations, or your workflows. This design decision — made early in the project's life — has turned out to be one of its most strategically important features, especially in China.
While the existing blog post on OpenClaw's China expansion covered the market growth story, this article dives into the technical side: which Chinese LLMs are integrated, how they connect, and what platforms have been built on top.
Alibaba Qwen: Official Documentation-Level Support
Alibaba Cloud's Model Studio has published official documentation for using Qwen models with OpenClaw. This isn't a community hack — it's a first-party integration guide hosted on alibabacloud.com.
Qwen (通义千问) is Alibaba's flagship LLM family. The integration works through the OpenAI-compatible API endpoint that Alibaba Cloud provides, which means OpenClaw treats it like any other OpenAI-compatible provider. Configuration is straightforward: point your OpenClaw instance at the Alibaba Cloud endpoint, provide your API key, and select your Qwen model variant.
This matters because Alibaba Cloud also offers OpenClaw hosting across 19 regions starting at $4/month. A user in China can spin up an OpenClaw instance on Alibaba Cloud, powered by Qwen, with Chinese messaging integrations — a fully domestic stack with no dependency on foreign AI providers or infrastructure.
Moonshot AI and Kimi Claw: Building a Platform on OpenClaw
Moonshot AI (月之暗面) has gone further than any other Chinese AI company. They haven't just integrated their Kimi models with OpenClaw — they've built an entire platform around it.
Kimi Claw is a dedicated AI agent platform built on the OpenClaw orchestration standard. It launched with:
- •Native support for Kimi models as the reasoning backend
- •A library of 5,000+ community skills
- •A managed hosting environment so users don't need to self-host
- •Integration with Chinese messaging platforms
There's also an official `kimi-integration` skill in the OpenClaw skills library, providing a step-by-step guide for integrating Moonshot AI's Kimi and Kimi Code models into any OpenClaw instance.
The Moonshot China Expansion was significant enough to be called out as a named feature in the OpenClaw v2026.2.3 release notes, alongside Cloudflare AI Gateway support. That's a signal of how important the Chinese market has become to the core project.
For local deployment, guides exist for running OpenClaw with Ollama and Kimi K2.5, giving users a fully offline option.
MiniMax M2.1
MiniMax, the Chinese AI company behind the Talkie app, has documented integration with OpenClaw through their M2.1 model. The integration follows the same OpenAI-compatible API pattern, making it a drop-in replacement for users who want to experiment with different model characteristics.
Feishu/Lark: The First Official Chinese Chat Platform
OpenClaw v2026.2.2 introduced official Feishu (飞书) / Lark support — described in the release notes as "the platform's debut as the first Chinese chat client." This was a milestone: the first time a Chinese messaging platform received first-party support in the core project, not just a community plugin.
The official `feishu-bridge` skill connects a Feishu bot to OpenClaw via WebSocket long-connection. Key detail: no public server, domain, or ngrok required. The bridge initiates an outbound connection from your OpenClaw instance to Feishu's servers, which means it works behind NAT and firewalls — a practical necessity in Chinese network environments.
DingTalk and WeCom: Community Connectors
For Alibaba's DingTalk (钉钉) and Tencent's WeCom (企业微信), the community has stepped in:
- •DingTalk connector by `xiaomingx`: Bridges DingTalk outgoing webhook messages to the OpenClaw Gateway. Available as a skill on the official registry.
- •WeCom plugin by `sunnoy` (`openclaw-plugin-wecom`): An Enterprise WeChat AI Bot plugin supporting streaming output, dynamic agent management, group chat integration, and command whitelisting.
- •Feishu connector by `m1heng` (`clawdbot-feishu`): An alternative community-built Feishu integration.
These connectors mean that the three major Chinese enterprise messaging platforms — Feishu, DingTalk, and WeCom — all have working OpenClaw integrations, covering the vast majority of Chinese workplace communication.
The Chinese Fork Ecosystem
Multiple community forks provide optimized experiences for Chinese users:
- •openclaw-cn (by jiulingyun): Chinese version with built-in Feishu integration and domestic network environment optimization — likely addressing connectivity issues related to the Great Firewall.
- •openclaw-cn (by AI-ZiMo): Another Chinese fork, regularly synced with the upstream project.
- •OpenClaw CN on SourceForge: Described as "focused on making a powerful open-source agent framework usable and understandable for Chinese-speaking developers."
These forks handle the practical realities of running OpenClaw in China: network routing, API endpoint selection, default model configuration, and UI localization.
The Architecture Advantage
What makes all of this possible is OpenClaw's clean separation between the orchestration layer and the model layer. The project doesn't assume you're using OpenAI. Any provider that exposes an OpenAI-compatible API — and most Chinese LLM providers do — works as a drop-in backend.
This is a deliberate architectural choice, and it's paying dividends. While other AI agent frameworks are tightly coupled to specific providers, OpenClaw's flexibility has made it the default orchestration layer for China's diverse LLM ecosystem.
What's Next
The trajectory is clear: Chinese AI companies aren't just using OpenClaw — they're building on it. Moonshot AI's Kimi Claw platform is the most ambitious example, but Alibaba's official documentation and Baidu's integration into their 700-million-user search app suggest this is an industry-wide trend.
As the foundation transition unfolds, one of the key questions will be how the governance structure balances the needs of the global community with the specific requirements of the Chinese market — different LLM providers, different messaging platforms, different network constraints, and different regulatory environments.