Can test-time scaling work for diffusion language models? In our #ICML2026# paper "UnMaskFork," we show that having multiple masked diffusion language models collaborate on a single answer improves performance on coding and math tasks. Blog: Test-time scaling is an actively researched technique that boosts LLM performance by using inference-time compute, for example, by having a model think longer or repeatedly refine its answers. This allows us to enhance performance simply by increasing computation during inference without relying on additional training, giving us the flexibility to balance compute costs and performance based on the specific use case. Unlike standard LLMs that generate text left-to-right, masked diffusion language models (MDLMs) generate text by gradually filling in a fully masked sequence. MDLMs can generate multiple parts of a sequence in parallel, offering potential speed-ups, and they can generate flexibly while seeing the entire sequence at once. This makes them an actively studied new paradigm in language modeling. We found that the standard LLM approach of "raising the temperature to increase randomness and generate diverse answers" does not work well for MDLMs like Dream-Coder. Instead of relying on this randomness, our proposed method, UnMaskFork (UMF), creates diversity through "model switching." Multiple MDLMs share the task of unmasking a single answer, and we use Monte Carlo Tree Search to search for a promising sequence in which different models handle different stages. Each model picks up where the others left off, filling in the parts it is most confident about. This collaborative approach allows us to explore diverse answers while maintaining generation quality, consistently outperforming existing test-time scaling methods on coding benchmarks and scaling effectively on math as well. Test-time scaling is also crucial for advancing MDLMs, and our work shows that UMF can sidestep the difficulties specific to them. UMF requires no additional training or changes to the models; it works simply by combining pre-trained models at inference time. This allows us to leverage the diversity of diffusion language models trained on different data and with different methods to improve performance. We believe the value of UMF will only grow as more diverse MDLMs emerge. This work is part of our broader research into "collective intelligence of AI," alongside methods like AB-MCTS and Sakana Fugu that have multiple LLMs collaborate. We'll continue pursuing research that turns model diversity into a source of strength. For details of the algorithm and illustrative examples showing how this collaboration works, please see our blog and paper. Paper: 🐟

Congratulations to the @GoogleDeepMind authors of "Asynchronous Methods for Deep Reinforcement Learning", recipient of the #ICML2026# Test of Time Award. This work shows that asynchronous actor-critic succeeds on a wide variety of continuous motor control problems as well as on a new task of navigating random 3D mazes using a visual input.

Learn more about how Isomorphic Labs' Drug Design Engine is using artificial intelligence to help solve disease. Join Agnieszka Podsiadlo and Andreas Loukas today at 3:00pm at the Google booth (#B206#) to discuss AI healthcare pipelines. #ICML2026#

Join Vahab Mirrokni and Michael Galkin at 12:30pm today at the Google booth (#B206#) for a session on graph foundation models. Learn how generalizable graph learning improves systems and coding, driving 20% datacenter efficiency gains and optimizing database queries. #ICML2026#

How can we map deep research ideas into practical robotics? Join Krzysztof Choromanski today at 11:00am at the Google booth (#B206#) to explore Gemini Robotics models, 3D Transformers, and object detection techniques. #ICML2026#

Pluralis v0.1 is a novel multimodal, multi-regional, and multilingual dataset built from a culture-first perspective. Meet Lora Aroyo today at 9:30am at the Google booth (#B206#) to explore localized safety evaluation paradigms. #ICML2026#

Co-Director is a hierarchical multi-agent framework that solves the challenge of narrative drift in long-video generation. Meet Yale Song and Yiwen Song today at 3:00pm at the Google booth (#B206#) to see generative storytelling in action. #ICML2026# Watch more here!

Coral NPU provides a glimpse of the future of LLMs on low-power platforms. See English-to-target language translation running on a low-power Torq-Coral NPU with Gregory Kielian today at 12:30pm at the Google booth (#B206#). #ICML2026#

Meet Toby Staines & Michal Kazmierski today at 11:30am at the Google booth (#B206#) to learn about AlphaEarth Foundations. This embedding field model yields a unified representation of the Earth's surface by assimilating multiple remote sensing data sources. #ICML2026#