Top Stories
4 outlets·5 reports·over 10h

Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model

ThinkingNews Desk · how this was written

Moonshot AI, a Chinese startup, released Kimi K2.5, an open-source model trained on roughly 15 trillion mixed visual and text tokens. The model processes text, images and video simultaneously, includes a coding agent, and can self-direct a swarm of up to 100 sub-agents, outperforming open-source peers in several tests.

Written from all 4 reports below, not from any single one.

How it was reported

  1. Hacker News·
    Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model

    Kimi released Kimi K2.5, an open-source visual SOTA-agentic model. The model is designed for advanced applications. Kimi K2.5 offers improved capabilities.

  2. TechMeme·
    Chinese startup Moonshot updates its Kimi model, saying K2.5 can process text, images, and videos simultaneously and beats its open-source peers in some tests (Bloomberg)

    Moonshot AI's updated Kimi model, K2.5, can process text, images, and videos simultaneously. It outperforms open-source peers in some tests. Alibaba Group Holding Ltd.-backed Moonshot AI released this upgrade.

  3. TechMeme·
    Moonshot says Kimi K2.5 builds on K2 with "pretraining over ~15T mixed visual and text tokens" and "can self-direct an agent swarm with up to 100 sub-agents" (Kimi)

    Kimi K2.5 builds on K2 with pretraining over 15T mixed visual and text tokens. It can self-direct an agent swarm with up to 100 sub-agents. Kimi K2.5 is the most powerful open-source model to date.

  4. TechCrunch·
    China’s Moonshot releases a new open-source model Kimi K2.5 and a coding agent

    China's Moonshot released Kimi K2.5, an open-source model trained on 15 trillion mixed visual and text tokens. The model is accompanied by a coding agent. This release aims to advance AI capabilities.

  5. VentureBeat·
    Moonshot AI debuts Kimi K2.5, most powerful open source LLM beating Opus 4.5 with swarm of parallel agents

    Moonshot AI debuts Kimi K2.5, a powerful open-source LLM. Kimi K2.5 supports visual and text inputs, outperforming models like Opus 4.5. It features Agent Swarm orchestration, allowing for parallel workflows and efficient scaling.

Related stories

Share: