TLDR
- Moonshot has launched its new AI model Kimi K2.5 with advanced multimodal capabilities.
- The model can process text, images, and video within a single unified system.
- Moonshot received $500 million in funding from investors including Alibaba and IDG Capital.
- The company is now targeting a new valuation of $5 billion through additional funding rounds.
- DeepSeek is expected to announce a new AI model after sharing research and code online.
- Zhipu and MiniMax raised over $1 billion through IPOs on the Hong Kong Stock Exchange.
Moonshot has introduced its new AI model, Kimi K2.5, with native multimodal capabilities, as competition intensifies in China. The launch follows Moonshot’s $500 million funding round that pushed its valuation to $4.3 billion. The announcement arrives as rivals DeepSeek, Zhipu, and MiniMax move forward with their own large-scale AI initiatives.
Kimi K2.5 Launches with Multimodal Features
Moonshot released Kimi K2.5 with a system that can process text, images, and video using a single architecture. The company confirmed that this is its most advanced model so far and expects it to close performance gaps. Its team said, “We believe K2.5 will stand strong in both proprietary and open-source model comparisons.”
Kimi K2.5’s unveiling signals Moonshot’s push to maintain leadership in China’s language model race. The model introduces enhancements designed for complex reasoning and high-level coding tasks. It also lays the groundwork for new enterprise AI tools.
The company stated that its automated coding assistant, which is in development, will directly compete with Anthropic’s Claude Code tool. This marks Moonshot’s entry into AI coding agents with enhanced features for software development. With Kimi K2.5, the startup intends to address both consumer and enterprise use cases.
DeepSeek Prepares Countermove as Rivals Scale Up
DeepSeek has not yet launched its upcoming model, but sources suggest an announcement is near. The company’s research lab has shared new publications and code on GitHub, signaling a major release. CEO Liang Wenfeng’s contributions have gained attention in China’s AI sector.
Analysts expect DeepSeek to respond soon, as competitors escalate efforts in model development and commercialization. Smaller firms have faced pressure as DeepSeek’s earlier R1 model achieved key technical milestones. However, DeepSeek has remained quiet on specific product details.
Meanwhile, Zhipu and MiniMax completed IPOs in Hong Kong and raised over $1 billion collectively in January. Their success has pushed other AI startups to secure funding and accelerate development. Competition continues to rise as model performance and commercialization take center stage.
Alibaba-Backed Funding Fuels Expansion and Valuation Push
Moonshot secured $500 million in funding in December with backing from Alibaba and IDG Capital, bringing its valuation to $4.3 billion. The company has now begun new funding rounds seeking a $5 billion valuation. These developments reflect investor confidence in Moonshot’s technology roadmap.
Alibaba’s support has aligned with recent model releases by its own AI division, including Qwen3-Max with reasoning capabilities. In response, Moonshot claims Kimi K2.5 reduces coding performance gaps against proprietary models. The company also maintains that its tools will deliver scalable solutions.
Zhipu recently launched GLM-Image, an AI model fully trained on domestic chips, bypassing foreign hardware dependency. Moonshot, Zhipu, and MiniMax now lead China’s top AI companies, once described under the label “War of One Hundred Models.” Their competition signals growing interest in AI innovation from investors and enterprises alike.
