Xiaomi Unveils Trio of Large AI Models in $8.7 Billion Bet
00:00/00:00
Listen to this article
1x
Subscribe now
Listen to the full version
Listen to the full version
Xiaomi launched three self-developed large models on March 19. Photo: VCG
Xiaomi Corp. on Thursday unveiled three in-house artificial intelligence large models, marking an aggressive push by the Chinese tech company to compete in the fast-evolving generative AI race.
The company introduced its flagship base model, Xiaomi MiMo-V2-Pro; a multimodal base model, Xiaomi MiMo-V2-Omni; and a speech synthesis model, Xiaomi MiMo-V2-TTS. The models are designed for a range of applications, including AI agent reasoning, multimodal processing and voice interaction.
You've accessed an article available only to subscribers
VIEW OPTIONS
Unlock exclusive discounts with a Caixin group subscription — ideal for teams and organizations.
Subscribe to both Caixin Global and The Wall Street Journal — for the price of one.
- RELATED
Share this article
Open WeChat and scan the QR code
- DIGEST HUB
Digest Hub
Back
Explore the story in 30 seconds
- Xiaomi launched three new large AI models—MiMo-V2-Pro, MiMo-V2-Omni, and MiMo-V2-TTS—covering advanced reasoning, multimodal processing, and speech synthesis.
- Backed by a planned 60 billion yuan ($8.7 billion) AI investment over three years, Xiaomi aims to be a global top-tier model developer, with over 16 billion yuan spent in 2024.
- MiMo-V2-Pro ranks seventh globally in benchmarks, and Xiaomi is integrating these models into its devices and AI agent ecosystem.
AI generated, for reference only
Who’s Who
- Xiaomi Corp.
- Xiaomi Corp. unveiled three in-house AI large models, including its flagship MiMo-V2-Pro, to compete in the generative AI race. With a planned investment of 60 billion yuan ($8.7 billion), Xiaomi aims to integrate advanced AI into its devices. The company's AI models have shown strong performance in testing and are priced competitively. Xiaomi also plans to integrate these models into its "human-car-home" ecosystem.
- Alibaba Group Holding Ltd.
- Alibaba Group Holding Ltd. is mentioned as the previous employer of Luo Fuli, who now leads Xiaomi's large-model team. Luo Fuli worked at Alibaba's DAMO Academy before joining Xiaomi in November 2025.
- DeepSeek
- DeepSeek is an organization where Luo Fuli, the team leader for Xiaomi's large-model development, previously worked. Luo Fuli contributed to the development of DeepSeek-V2 during her tenure there.
- OpenRouter
- OpenRouter is an API aggregation platform where Xiaomi's MiMo-V2-Pro and MiMo-V2-Omni models were anonymously tested as "Hunter Alpha" and "Healer Alpha" prior to their official launch. These models topped daily usage rankings, generating over 1 trillion token calls and attracting significant developer interest on the platform.
- Artificial Analysis
- Artificial Analysis is a third-party evaluator that assessed Xiaomi's MiMo-V2-Pro model. It ranked MiMo-V2-Pro seventh globally and third among Chinese models after testing it across 10 benchmark tests, where it scored 49.
- MiniMax
- MiniMax M2.7 is a large language model based in China. It scored 50 in Artificial Analysis's benchmark tests, tying with GLM-5 and ranking second among Chinese models, surpassed only by Gemini 3.1 Pro Preview and GPT-5.4 globally. This places it above Xiaomi's MiMo-V2-Pro model, which scored 49.
AI generated, for reference only
What Happened When
- November 2025：
- Luo Fuli joined Xiaomi as team leader for the large-model team.
- December 2025：
- Xiaomi released its first open-source large language model, MiMo-V2-Flash.
- Earlier in March 2026：
- Xiaomi began limited testing of the AI agent Xiaomi miclaw, built on MiMo models.
- March 19, 2026：
- Xiaomi officially unveiled three in-house AI large models: MiMo-V2-Pro, MiMo-V2-Omni, and MiMo-V2-TTS.
- By March 19, 2026：
- Xiaomi's AI-related spending in 2026 exceeded 16 billion yuan.
AI generated, for reference only
- PODCAST
- MOST POPULAR
Sign in with Google
Sign in with Facebook
Sign in with 财新