Arian Mousakhan

🎓 PhD @ Uni Freiburg · Computer Vision Lab 🌍 Researching World Models & Generative AI

[Arian Mousakhan @Arianmousa]

Orbis 2 is accepted at #NeurIPS2026! 🚗
A hierarchical driving world model.
Code, models & demo 👇
https://t.co/zRZFf6rsNG

[Michael Black @Michael_J_Black]

What is the role of academic computer vision research in the age of increasingly powerful large models? Is GPT-6 Astra a step change? How can a researcher have an impact today in academia?

These are the questions I ask myself as I head off to ECCV 2026, a conference I’ve attended since 1992. One of my papers this year is VIGA, a method that takes an image as input and outputs a 3D Blender scene that represents that image. This is a classical inverse-graphics task and VIGA was the first method to solve it using an agentic approach.

The idea is now several years old and the first version of the paper was rejected. This delayed publication significantly. After it was accepted at ECCV, it was quickly surpassed by people using Claude Code for the same purpose. Today GPT-6 Astra blows away all previous results. But we still head off to ECCV to tell the community about our invention that is now fully out of date.

The way academic work often progresses is that one reads recent papers, notices that they have limitations, comes up with a new idea, explores this, publishes it, etc. Any published paper I read today is based on ideas that are at least a year old. And those ideas were based on the literature of the time, which was also a year old. That means that any paper I see at ECCV is likely two years out of date. In AI today, two years means your work is likely irrelevant.

At CVPR this summer I noticed that many authors have not gotten the message. They continue to work on “old” problems that have a long history. This history is based on assumptions about how the “vision problem” will be “solved”. The truth is that it is being solved in a very different way and many of these problems are no longer relevant. Another group of papers focuses on very niche problems where large models likely fail because of insufficient data or lack of business interest. The impactful papers were largely from industry and had long author lists and massive data+compute behind them. These papers were also out of data, describing systems that had been released months before, but at least they served to provide the community with more complete documentation and analysis of commercial systems.

So what should academics do? First, we need to put aside the tools we’ve used for years and start from scratch. Every project should start by trying really hard to solve the problem with existing tools. I would like to see every paper begin with a detailed experimental analysis of how existing models perform and why they fail (if they do). This gives the kind of insight we need today. Then, assuming current models fail, the solution should provide some fundamental insight that will outlive the next release of such models.

Reviewers today still focus on technical novelty. This pushes people to focus on tweaking architectures rather than clearly moving the field forward. Papers need to be judged based on their novel insight and not their novel technical contribution. This is a real shift in thinking but it focuses us on what matters - progress of the field.

If we want there to be a “field” of computer vision, then it can’t become a marginal backwater, focusing on esoteric problems. If you haven’t tried using Astra (or whatever comes next) to solve your problem, then you have not done your homework. This omission should be seen as negatively as not having a previous work section.

Concretely, I think papers should include a new section analogous to “Related Work” where that related work is current models and how they perform on the task. Reviewers should start asking for this and expecting authors to be able to articulate their insights about the limitations of existing large models.

I'm interested in your thoughts.

[Arian Mousakhan @Arianmousa]

@Jesvinquotes Secular!

[Arian Mousakhan @Arianmousa]

@moshhamedani Although I would prefer: “We arrested him, and Iranian patriots killed him like a rat.” 😂

[Arian Mousakhan @Arianmousa]

@realDonaldTrump MIGA ❤️

[Arian Mousakhan @Arianmousa]

@stats_feed Khamenei must be added

[Arian Mousakhan @Arianmousa]

@TrumpTruthOnX ❤️🤍💚

[Arian Mousakhan @Arianmousa]

@lexfridman Thanks for your support.

[Arian Mousakhan @Arianmousa]

Orbis: https://t.co/TIEQV68sa6

[Arian Mousakhan @Arianmousa]

@karim_farid0 and I are heading to #EurIPS in Copenhagen. We will present our work, Orbis, on Driving World Model. Find us on Wednesday at poster session. Would be glad to talk about World Model, representation learning, self supervised learning.

[Arian Mousakhan @Arianmousa]

A great talk from Yann LeCun explaining World Model. We are also steering Orbis toward more representation learning and analysing its performance on downstream tasks.
Orbis: https://t.co/96zreXrQ21

Yann LeCun talk: https://t.co/0IzuAqqmIK

[Arian Mousakhan @Arianmousa]

The Orbis code and checkpoints are now available!
Explore, experiment, and feel free to reach out with any questions or discussions.

📄 https://t.co/qer0Wv7J3i
🎬 https://t.co/96zreXrQ21
💻https://t.co/IojbH7szxo
🤗 https://t.co/2x4V0vCm7U

[Arian Mousakhan @Arianmousa]

Compact & scalable: 469M params, 280h video. SOTA metrics and realistic trajectories in tough urban/turning scenes. #WorldModels #AutonomousDriving #GenerativeAI

[Arian Mousakhan @Arianmousa]

Orbis is in #NeurIPS2025! 🎉 a continuous autoregressive driving world model with a hybrid tokenizer to compare apples-to-apples vs. discrete tokens. Result: continuous wins—more robust, longer rollouts, better turns. Read: https://t.co/oYjsw2YteS: https://t.co/KsQpYSc6ve

[Arian Mousakhan @Arianmousa]

Orbis can generate realistic and diverse scenes from just 5 context frames.