The race in the large model field has entered a fiercely competitive phase in the second half of 2026. In addition to the official release of DeepSeek V4, the new generation large model Kimi K3 from Moonshot AI, a unicorn in the field of artificial intelligence, has also been confirmed to be released within this month.

Although the official has not yet disclosed the specific release date and detailed technical documents of K3, according to insiders, this new model shows a significant improvement in performance. It is expected to have a parameter scale of up to 2.5 trillion, far exceeding the 1.6 trillion parameters of DeepSeek V4 Pro, and surpassing Baidu's WENXIN 5.0 (2.4 trillion), and is expected to become the largest "leader" in terms of parameter scale among domestic large models.

In terms of functionality, Kimi K3 is expected to integrate a super-long context window of 1 million and advanced multimodal processing capabilities. This combination of "massive parameters + all-round multimodal" is seen as a key factor for domestic models to challenge global top-level standards. Previously, Moonshot AI had continuously launched Kimi K2.7 Code and its high-speed version in June, maintaining a frequent update cycle of "one update per month." The release of K3 will undoubtedly further solidify its first-mover advantage in the technology ecosystem.

Meanwhile, the large model market in July is a battleground for many competitors. On the international side, Google's Gemini 3.5 Pro, originally scheduled to launch in June, has been postponed to this month; the Grok series under Musk's company is also active frequently. With its deep integration with the AI coding tool Cursor, the Grok v9 version, which is expected to match Opus's performance, is expected to be completed by the end of July and planned to be officially launched in August.

As model manufacturers continue to invest in underlying algorithms and commercial applications, the competition landscape of artificial intelligence in the second half of the year will become increasingly clear. For developers and enterprise users, the dual breakthroughs of domestic large models in parameter scale and multimodal capabilities mean that there will be more high-performance, cost-effective professional tools available in the future. The industry generally expects that the launch of Kimi K3 will bring unexpected performance improvements to the domestic large model group, leading the industry to a new level of intelligence.