%20--%3e%3cdefs%3e%3cstyle%3e%20.st0%20{%20fill:%20%23061b40;%20}%20.st1%20{%20fill:%20%23306af1;%20}%20.st2%20{%20fill:%20%235ce5cf;%20}%20%3c/style%3e%3c/defs%3e%3cg%3e%3cpath%20class='st0'%20d='M55,10.5h9v3h-9v9h12V7.5h-12v3ZM64,19.5h-6v-3h6v3Z'/%3e%3cpolygon%20class='st0'%20points='69%2016.5%2078%2016.5%2078%2019.5%2069%2019.5%2069%2022.5%2081%2022.5%2081%2013.5%2072%2013.5%2072%2010.5%2081%2010.5%2081%207.5%2069%207.5%2069%2016.5'/%3e%3cpolygon%20class='st0'%20points='95%2010.5%2095%207.5%2083%207.5%2083%2022.5%2095%2022.5%2095%2019.5%2086%2019.5%2086%2016.5%2095%2016.5%2095%2013.5%2086%2013.5%2086%2010.5%2095%2010.5'/%3e%3cpath%20class='st0'%20d='M40,1.5v21h11.6l1.4-1.4v-7.6h0c0,0-1.4-1.5-1.4-1.5l1.4-1.4V2.9l-1.4-1.4h-11.6ZM50,19.5h-7v-6h7v6ZM50,10.5h-7v-6h7v6Z'/%3e%3c/g%3e%3cpath%20class='st1'%20d='M23.1,24L14.7,4.8l-4.9,11.2h3.8l-1.8,4H3.3L12.1,0H2C.9,0,0,.9,0,2v20c0,1.1.9,2,2,2h21.1Z'/%3e%3cpath%20class='st2'%20d='M34,0h-16.8l10.6,24h6.2c1.1,0,2-.9,2-2V2C36,.9,35.1,0,34,0ZM32.5,20h-4V4h4v16Z'/%3e%3c/svg%3e)
The latest intelligence indicates that Kimi K2.5 has quietly been released online, and this upgrade has quickly sparked industry discussions.
According to multiple sources, Kimi K2.5 was recently launched on the Moonshot AI official website (kimi.com) in a low-key manner. This version was not widely advertised but was silently pushed to users through the web interface. User feedback shows that the original K2 model has automatically switched to K2.5 in the chat interface, including the thinking mode and multimodal support. This update is seen as an iterative optimization of the K2 series, aiming to further enhance the AI's intelligent interaction capabilities. AIbase believes that this "silent rollout" strategy may be to rapidly iterate and collect user feedback, avoiding early exposure of potential issues.
New Feature Highlights: Dual Upgrade in Visual and Tool Usage
The biggest highlight of Kimi K2.5 is its native support for visual capabilities and tool usage. Unlike previous versions, users can now directly upload images for analysis, such as generating 3D models based on images or interpreting complex layouts. Testing shows that the model performs exceptionally well in visual tasks, accurately handling tasks like converting TV scene floor plans or generating Three.js format 3D models. In addition, the tool usage feature makes Kimi K2.5 closer to "agent-like intelligence," capable of step-by-step reasoning to solve math, logic, or programming problems. Compared to K2, the quality improvement from K2 to K2.5 is described as "a huge leap similar to the upgrade from Gemini 2.5 Pro to Gemini 3 Pro," setting higher standards for future K3 versions. AIbase analyzes that these enhancements will significantly improve Kimi's practicality in real-world applications, especially within the open-source community and developer groups.
User Feedback: Outstanding Performance, High Expectations for Open Source
From online feedback, Kimi K2.5 has received praise since its release. Multiple users stated that the model easily passed coding and visual tests, such as quickly generating a 3D model of a friend's apartment or demonstrating stronger reasoning capabilities when dealing with complex prompts. The open-source community is particularly excited, with some praising it as "a joyful day for the open-source community" and expecting Moonshot AI to open-source the full model soon. Currently, some features, including thinking and multimodal support, can already be accessed via API. AIbase noticed that although the open-source announcement has not been officially made, related discussions suggest this could become the next milestone for the Kimi series, and user recognition of performance has injected new vitality into the AI competition.
In summary, the release of Kimi K2.5 marks Moonshot AI's continued efforts in the AI field. This upgrade not only strengthens the core competitiveness of the model but also provides users with a smarter and more comprehensive tool. AIbase will continue to track subsequent developments, so stay tuned for more AI front-line reports.