Home
Moonshot AI’s Kimi K2.5, the first trillion-parameter multimodal model, retires at month’s end
Moonshot announced via Kimi’s official Weibo that its inaugural trillion-parameter multimodal model, Kimi K2.5, will be retired by the end of this month. Launched and open-sourced in January, K2.5 stands as Kimi’s most versatile model to date, featuring a native multimodal architecture that handles visual and text inputs, supports both thinking and non-thinking modes, and enables dialogue and Agent tasks.

K3 Takes Over, Parameter Scale Rises Again
K2.5 achieved state-of-the-art performance in open-source models on tasks such as Agents, code, images, videos, and general intelligence. It is positioned as a general-purpose multimodal model that can switch between dialogue, complex reasoning, and automation tasks.
Its retirement also marks that Moonshot has officially shifted its focus to Kimi K3, which was open-sourced in July. This 2.8 trillion parameter model is built using the KDA hybrid linear attention mechanism and attention residual technology, and it natively supports visual understanding with a context window of up to 1 million tokens. The official stated that K3 is the first open-source model to reach a scale of 2.8 trillion parameters, and for the past 12 months, Kimi has held the record for the largest open-source model scale for 9 of those months.
Related article
AI founders weigh open vs. closed models at TechCrunch Disrupt 2026
For AI startups, selecting a model is no longer just a one-time architectural choice.Open models are getting better. Frontier APIs keep improving. Models can be tailored for specific tasks, and some companies are building products that use multiple m
How AI and Robotics Are Transforming bp’s Oil Production
Digital twin technology enables bp to simulate energy asset operations, identifying improvement opportunities and predicting failures before they occur. Credit: bpRicky Burns, Transformation & Technology Manager at bp, explains how digital tools help
How to create free AI talking avatars?
Modern AI systems rely heavily on up-to-date, structured, and market-specific web data. According to McKinsey’s 2025 global survey, 88% of respondents reported that their organizations regularly utilize AI in at least one business function. As enterp
Related Special Topic Recommendations
Comments (0)
0/500
Moonshot announced via Kimi’s official Weibo that its inaugural trillion-parameter multimodal model, Kimi K2.5, will be retired by the end of this month. Launched and open-sourced in January, K2.5 stands as Kimi’s most versatile model to date, featuring a native multimodal architecture that handles visual and text inputs, supports both thinking and non-thinking modes, and enables dialogue and Agent tasks.

K3 Takes Over, Parameter Scale Rises Again
K2.5 achieved state-of-the-art performance in open-source models on tasks such as Agents, code, images, videos, and general intelligence. It is positioned as a general-purpose multimodal model that can switch between dialogue, complex reasoning, and automation tasks.
Its retirement also marks that Moonshot has officially shifted its focus to Kimi K3, which was open-sourced in July. This 2.8 trillion parameter model is built using the KDA hybrid linear attention mechanism and attention residual technology, and it natively supports visual understanding with a context window of up to 1 million tokens. The official stated that K3 is the first open-source model to reach a scale of 2.8 trillion parameters, and for the past 12 months, Kimi has held the record for the largest open-source model scale for 9 of those months.
AI founders weigh open vs. closed models at TechCrunch Disrupt 2026
For AI startups, selecting a model is no longer just a one-time architectural choice.Open models are getting better. Frontier APIs keep improving. Models can be tailored for specific tasks, and some companies are building products that use multiple m
How AI and Robotics Are Transforming bp’s Oil Production
Digital twin technology enables bp to simulate energy asset operations, identifying improvement opportunities and predicting failures before they occur. Credit: bpRicky Burns, Transformation & Technology Manager at bp, explains how digital tools help
How to create free AI talking avatars?
Modern AI systems rely heavily on up-to-date, structured, and market-specific web data. According to McKinsey’s 2025 global survey, 88% of respondents reported that their organizations regularly utilize AI in at least one business function. As enterp











