When Claude Leaves China: How Domestic AI Models Are Rising to Fill the Gap

Anthropic recently announced that any company controlled by Chinese capital, regardless of its registration location, can no longer use Claude or related services such as Claude Code, dealing a direct blow to Chinese enterprises that rely on these tools for R&D and content creation.

Faced with the risk of supply cuts, developers must ask how to continue their work without Claude and which domestic AI models can serve as reliable replacements while opening new opportunities.

01 Why Claude Became So Popular

1. Exceptional comprehension of complex, multi‑step instructions

Claude can follow intricate, layered commands, breaking tasks into steps and delivering coherent results, much like a top‑performing student who handles every requirement.

2. Structured output for easy data integration

The model can generate JSON and other structured formats directly, saving developers from tedious data cleaning and transformation.

3. Strong coding capabilities with Claude Code

Claude Code acts as an AI programming agent, understanding intent, writing and completing code, fixing bugs, and even automating end‑to‑end code production.

4. Emphasis on safety and controllability

Anthropic prioritises reducing harmful outputs and hallucinations, offering a more trustworthy AI experience.

02 Domestic AI Models Gaining Momentum

China’s AI model market is exploding, projected to surpass a trillion‑yuan valuation, with notable models such as Baidu’s Wenxin, Alibaba’s Tongyi, ByteDance’s Doubao, and Zhipu AI’s GLM series.

In the programming arena, several home‑grown models stand out:

Alibaba Qwen3‑Coder : an open‑source coding model that excels at writing, completing, and debugging code, dramatically boosting developer productivity.

Alibaba Tongyi Lingma : a comprehensive coding assistant offering multi‑file edits and an “AI coding agent” experience.

Zhipu GLM‑4.5 : a “foundation model for agents” that combines reasoning, coding, and agent capabilities, achieving state‑of‑the‑art performance on benchmark tests.

These models are already being integrated into development pipelines to automate testing, deployment, and documentation, freeing engineers to focus on creative problem‑solving.

03 Where Domestic Models Still Lag Behind Claude Code

1. Maturity of the programming‑agent mode

Claude Code provides end‑to‑end automation from natural‑language task description to code generation, testing, and deployment, whereas most Chinese models remain auxiliary tools rather than full agents.

2. Context handling capacity

Claude excels at long‑context recall, enabling accurate code generation for large projects; domestic models are improving but still struggle with very long or complex contexts.

3. Ecosystem and community support

Claude benefits from a large, active developer community and a rich ecosystem of plugins and integrations, which Chinese models are still building.

4. Safety and compliance

While Anthropic’s safety work is notable, the abrupt restriction highlights the need for domestic models to ensure stable, compliant services that cannot be suddenly withdrawn.

Overall, Claude’s departure is both a setback and a catalyst, urging Chinese AI developers to accelerate innovation, close performance gaps, and achieve true independence.