--- title: Top 5 Chinese LLMs: The Models Powering China’s AI Surge... description: A roundup of five leading Chinese large language models shaping AI from 2024 into 2026: DeepSeek R1 (671B MoE, MIT-licensed, strong on math/coding), Alibaba... canonical: https://daily.dev/posts/top-5-chinese-llms-the-models-powering-china-s-ai-surge-in-2024-25-333vub8d0 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Top 5 Chinese LLMs: The Models Powering China’s AI Surge in 2024–25 | daily.dev og:description: A roundup of five leading Chinese large language models shaping AI from 2024 into 2026: DeepSeek R1 (671B MoE, MIT-licensed, strong on math/coding), Alibaba... og:url: https://daily.dev/posts/top-5-chinese-llms-the-models-powering-china-s-ai-surge-in-2024-25-333vub8d0 og:image: https://api.daily.dev/og/posts/333vUB8D0.png og:image:alt: Top 5 Chinese LLMs: The Models Powering China’s AI Surge in 2024–25 og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Top 5 Chinese LLMs: The Models Powering China’s AI Surge in 2024–25 **[PromptLayer Blog](https://daily.dev/sources/promptlayer)** · 5 min read · 1 upvotes · 1 comments ## Summary A roundup of five leading Chinese large language models shaping AI from 2024 into 2026: DeepSeek R1 (671B MoE, MIT-licensed, strong on math/coding), Alibaba Qwen-3 (235B MoE with hybrid thinking/fast modes), Baidu ERNIE 4.5 and X1 (multimodal and agentic), Huawei PanGu-Σ/5.0 (trillion-parameter industrial suite), and Zhipu GLM-4.5 (agent-native, 355B MoE, #3 globally on benchmarks). Key themes include Mixture-of-Experts efficiency, agentic tool use, open-source licensing, and pricing far below Western equivalents. These models are closing the gap with GPT-4 on reasoning, coding, and multimodality benchmarks. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How many parameters does DeepSeek R1 activate per query and what is its architecture? DeepSeek R1 uses a 671-billion-parameter Mixture-of-Experts architecture but activates only 37 billion parameters per query for efficiency. It is MIT-licensed, available as downloadable weights from 1.5B to 70B, and trained heavily with reinforcement learning for math and coding, scoring 79.8% pass@1 on AIME and an estimated Codeforces Elo of 2029. _Developers comparing efficient MoE reasoning models can track new open-source LLM releases on daily.dev._ ### What is Alibaba Qwen-3's hybrid reasoning mode and how do I use it? Qwen-3 lets developers toggle between an analytical thinking mode for hard problems and a fast mode for simple queries, switchable per-query using the /think flag. The flagship model has 235B total parameters with 22B active, was trained on 36 trillion tokens, natively supports Alibaba's Model Context Protocol for tool use, and is Apache 2.0 licensed via Alibaba Cloud Model Studio. _Teams building agentic workflows can follow hybrid-reasoning model updates like this on daily.dev._ ### How much cheaper is Baidu ERNIE 4.5 API pricing compared to OpenAI? Baidu ERNIE 4.5 API pricing is just ¥0.004 per 1,000 tokens, orders of magnitude cheaper than OpenAI's rates. The model is multimodal, natively processing text, images, and audio, uses FlashMask dynamic attention and Heterogeneous MoE, and Baidu has open-sourced a version with up to 424 billion parameters. _Developers weighing AI API costs against performance can keep tabs on pricing shifts like this via daily.dev._ ## Community discussion Top comments from developers on daily.dev. **@phatdangminh** · 0 upvotes > Many thanks! ## Similar posts on daily.dev - [Top 5 Chinese LLMs Compared: Technical Innovation and Strategic Advantages](https://daily.dev/posts/top-5-chinese-llms-compared-technical-innovation-and-strategic-advantages-omie73nyl) · PromptLayer Blog · 1 upvotes · 0 comments - [The Top 10 Large Language Models \(LLMs\) of 2025: The Age of Cognitive Giants](https://daily.dev/posts/the-top-10-large-language-models-llms-of-2025-the-age-of-cognitive-giants-aobcrkv0h) · C\# Corner · 0 upvotes · 0 comments - [2025: The year in LLMs](https://daily.dev/posts/2025-the-year-in-llms-t60jvoyxr) · Simon Willison · 4 upvotes · 0 comments - [Top 5 Agentic AI LLM Models](https://daily.dev/posts/top-5-agentic-ai-llm-models-lilaxsxxv) · Machine Learning Mastery · 1 upvotes · 0 comments --- Tags: [#open-source](https://daily.dev/tags/open-source), [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#deepseek](https://daily.dev/tags/deepseek), [#mixture-of-experts](https://daily.dev/tags/mixture-of-experts) [View this post on daily.dev](https://daily.dev/posts/top-5-chinese-llms-the-models-powering-china-s-ai-surge-in-2024-25-333vub8d0) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Top 5 Chinese LLMs: The Models Powering China’s AI Surge in 2024–25","url":"https://daily.dev/posts/top-5-chinese-llms-the-models-powering-china-s-ai-surge-in-2024-25-333vub8d0","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/top-5-chinese-llms-the-models-powering-china-s-ai-surge-in-2024-25-333vub8d0"},"datePublished":"2026-07-06T00:59:52.784Z","dateModified":"2026-09-14T08:13:27.064Z","description":"A roundup of five leading Chinese large language models shaping AI from 2024 into 2026: DeepSeek R1 (671B MoE, MIT-licensed, strong on math/coding), Alibaba...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/1dba426dbb47f077430e1cdc2a4b9492?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/1dba426dbb47f077430e1cdc2a4b9492?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"PromptLayer Blog","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"PromptLayer Blog","logo":"https://media.daily.dev/image/upload/s--YNkmMZnf--/f_auto,q_auto/v1780213213/logos/promptlayer?_a=BAMAMiWQ0","url":"https://daily.dev/sources/promptlayer"},"commentCount":1,"discussionUrl":"https://daily.dev/posts/top-5-chinese-llms-the-models-powering-china-s-ai-surge-in-2024-25-333vub8d0","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":1}],"keywords":"open-source,llm,ai-agents,deepseek,mixture-of-experts","timeRequired":"PT5M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"PromptLayer Blog","item":"https://daily.dev/sources/promptlayer"},{"@type":"ListItem","position":3,"name":"Top 5 Chinese LLMs: The Models Powering China’s AI Surge in 2024–25"}]} {"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/top-5-chinese-llms-the-models-powering-china-s-ai-surge-in-2024-25-333vub8d0","comment":[{"@type":"Comment","text":"Many thanks!","datePublished":"2026-07-06T05:54:38.745Z","url":"https://daily.dev/posts/333vUB8D0#c-KwesmXkyL","author":{"@type":"Person","name":"Phat Dang Minh","url":"https://daily.dev/phatdangminh","image":"https://media.daily.dev/image/upload/s--uEe-dzLM--/f_auto/v1747452091/avatars/avatar_U9YNpHvqD7jM3IxMsDTsP?_a=BAMClqUq0"}}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/top-5-chinese-llms-the-models-powering-china-s-ai-surge-in-2024-25-333vub8d0#faq","mainEntity":[{"@type":"Question","name":"How many parameters does DeepSeek R1 activate per query and what is its architecture?","acceptedAnswer":{"@type":"Answer","text":"DeepSeek R1 uses a 671-billion-parameter Mixture-of-Experts architecture but activates only 37 billion parameters per query for efficiency. It is MIT-licensed, available as downloadable weights from 1.5B to 70B, and trained heavily with reinforcement learning for math and coding, scoring 79.8% pass@1 on AIME and an estimated Codeforces Elo of 2029. Developers comparing efficient MoE reasoning models can track new open-source LLM releases on daily.dev."}},{"@type":"Question","name":"What is Alibaba Qwen-3's hybrid reasoning mode and how do I use it?","acceptedAnswer":{"@type":"Answer","text":"Qwen-3 lets developers toggle between an analytical thinking mode for hard problems and a fast mode for simple queries, switchable per-query using the /think flag. The flagship model has 235B total parameters with 22B active, was trained on 36 trillion tokens, natively supports Alibaba's Model Context Protocol for tool use, and is Apache 2.0 licensed via Alibaba Cloud Model Studio. Teams building agentic workflows can follow hybrid-reasoning model updates like this on daily.dev."}},{"@type":"Question","name":"How much cheaper is Baidu ERNIE 4.5 API pricing compared to OpenAI?","acceptedAnswer":{"@type":"Answer","text":"Baidu ERNIE 4.5 API pricing is just ¥0.004 per 1,000 tokens, orders of magnitude cheaper than OpenAI's rates. The model is multimodal, natively processing text, images, and audio, uses FlashMask dynamic attention and Heterogeneous MoE, and Baidu has open-sourced a version with up to 424 billion parameters. Developers weighing AI API costs against performance can keep tabs on pricing shifts like this via daily.dev."}}]} ```