Unverified60% confidenceFactExact time
Nearly all of China's most advanced large models follow the open-source route, while top US models such as those from OpenAI and Anthropic are without exception closed-source.
2
Sources
60%
Confidence
Medium-term (~90 days)
Relevance
7/22/2026
First Seen
Valid until: 10/20/2026
Sources
Related Claims
UnverifiedCompanies such as OpenAI, Anthropic, Google, and Meta continue to release increasingly powerful foundation models, while open-source models like Llama and Mistral have already substantially narrowed the capability gap with commercial models.71% similarUnverifiedDoubao is more convenient for Chinese localization ecosystem integration, while OpenAI models have greater advantages in cross-domain reasoning and long-form expression.69% similarUnverifiedChinese open-source models such as Qwen (Alibaba) and Yi (01.AI) have continuously refreshed the performance leaderboards for open-source models.67% similarUnverifiedCommon open-source embedding models include BGE, M3E, and text-embedding-ada-002, with significant performance differences among different models in Chinese-language scenarios.67% similarUnverifiedDomestic tools provide priority support and out-of-the-box integration for Chinese Embedding models (such as the BGE and M3E series) and mainstream domestic model APIs (Baidu Wenxin, Alibaba Tongyi, and Zhipu GLM).66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/590216API
curl https://kongchang.com/api/v1/knowledge/claims/590216MCP
get_claim(id=590216)