# Kimi K2.5, DeepMind's Project Genie, and the OpenClaw frenzy

> Week of 26 Jan 2026: Moonshot's multimodal Kimi K2.5, a playable world model from DeepMind, DeepSeek-OCR 2, and the open agent that went viral.

Published 13 Oct 2026. Coverage: 26 Jan 2026 to 1 Feb 2026. By Ali Asif.

## Items

1. **Kimi K2.5** (Moonshot AI, Open weights): A natively multimodal open model with an agent-swarm mode, from the lab behind Kimi K2. [Primary source](https://www.kimi.com/blog/kimi-k2-5)
2. **Project Genie** (Google DeepMind, Image, video, world): Genie 3 became something users can try: you describe a world and walk around in it. [Primary source](https://deepmind.google/blog/project-genie-experimenting-with-infinite-interactive-worlds/)
3. **DeepSeek-OCR 2** (DeepSeek, Open weights): A second version of DeepSeek's open document reader, which matters to anyone digitising paper-heavy workflows. [Primary source](https://huggingface.co/deepseek-ai/DeepSeek-OCR-2)
4. **Trinity Large Preview** (Arcee AI, Open weights): A large open mixture-of-experts model trained in the US, released as a preview. [Primary source](https://huggingface.co/arcee-ai/Trinity-Large-Preview)
5. **OpenClaw** (OpenClaw (open source), Agents and coding): A personal agent project that went viral and was renamed twice in a week, showing both the demand for agents and the security risk. [Primary source](https://github.com/openclaw/openclaw)

Four launches, one viral side project, and a lot of people trying to work out what to trust.

## Kimi K2.5

Moonshot released Kimi K2.5 on 27 January. It is multimodal from the ground up, so images and video go in alongside text, and it adds an agent-swarm mode where the model splits a task across parallel sub-agents. The weights are open.

Two details caught my eye. One is that vision is no longer a bolt-on in the open world. The other is the swarm idea: instead of one agent grinding through a long list, a coordinator hands pieces to workers. That is exactly how I structure my own agents, with one planner and several narrow workers, so it is good to see a lab build it into the model.

I would test it on a real document task before putting it in a workflow. Open does not mean good for your job, it means you are allowed to find out.

## Worlds you can walk in

Google DeepMind opened Project Genie to users, built on Genie 3. You write a prompt, it builds an interactive world, and you move through it in real time. The worlds are short-lived and rough, and DeepMind lists limits openly. As a research direction it is a serious one, because world models are one path toward training robots in simulation.

## Documents and a US open model

DeepSeek-OCR 2 improves the open reader released late last year. Arcee released a preview of Trinity Large, a big open mixture-of-experts model trained in the US, which gives Western teams another open option at the large end.

## The agent that went viral

The open-source personal agent that began as Clawdbot was renamed Moltbot and then OpenClaw within days, and its GitHub stars shot up. The attention shows how many people want an assistant that acts on their behalf. The warnings were just as loud: an agent wired into messages and accounts can be steered by hostile text.

Zooming out, the week shows the two speeds of this field. On one side you have releases with proper model cards and licences. On the other you have a hobby project that piles up GitHub stars and forces a rename twice. The second kind moves faster than any company can respond, which is exactly why I keep my own agents boring: narrow tools, written permissions, and no access to anything I would not hand a new intern.

**What I'd watch next:** the first proper security guidance for personal agents, since adoption is clearly outrunning it.

## My take

The OCR release is the one that touches my desk. Tender packs and supplier quotes arrive as scanned PDFs, and a good open reader is the first step of any procurement automation. OpenClaw is the cautionary tale next to it: a single agent with access to your messages, files and keys is a very attractive target, so I keep mine on narrow permissions.

## Sources

- https://www.kimi.com/blog/kimi-k2-5
- https://deepmind.google/blog/project-genie-experimenting-with-infinite-interactive-worlds/
- https://huggingface.co/deepseek-ai/DeepSeek-OCR-2
- https://huggingface.co/arcee-ai/Trinity-Large-Preview
- https://github.com/openclaw/openclaw

Canonical: https://heyaliasif.com/ai/2026-01-26-week-in-ai