We present Magma, a foundation model that serves multimodal AI agentic tasks in both the digital and physical worlds. Magma is a significant extension of vision-language (VL) models in that it not only retains the VL understanding ability (verbal intelligence) of the latter, but is also equipped with the ability to ground and act in the visual-spatial world (spatial-temporal intelligence). To endow agentic capabilities for tasks ranging from UI navigation to robot manipulation, Magma is trained on large amounts of heterogeneous datasets that span from images, videos to robotics data, where actionable visual objects (e.g. clickable buttons in GUI) in images are labeled by Set-of-Mark (SoM) for action grounding, and object movements (e.g. trace of human hands or robotic arms) in videos are labeled by Trace-of-Mark (ToM) for action planning. Extensive experiments show that SoM and ToM help bridge the gap between verbal and action abilities and significantly enhance spatio-temporal intelligence which is fundamental to agentic tasks, as shown in Fig. 1. In particular, Magma creates new state-of-the-art results on UI navigation and robotic manipulation tasks, outperforming previous models that are specifically tailored to these tasks. Moreover, Magma preserves strong multimodal understanding ability and compares favorably to popular large multimodal models that are trained on much larger datasets. We have made our model and code public for reproducibility1.

Spatial intelligence (SI) represents a cognitive ability encompassing the visualization, manipulation, and reasoning about spatial relationships, underpinning disciplines from neuroscience to robotics. We introduce SITE, a benchmark dataset towards SI Thorough Evaluation in a standardized format of multi-choice visual question-answering, designed to assess large vision-language models' spatial intelligence across diverse visual modalities (single-image, multi-image, and video) and SI factors (figural to environmental scales, spatial visualization and orientation, intrinsic and extrinsic, static and dynamic). Our approach to curating the benchmark combines a bottom-up survey about 31 existing datasets and a top-down strategy drawing upon three classification systems in cognitive science, which prompt us to design two novel types of tasks about view-taking and dynamic scenes. Extensive experiments reveal that leading models fall behind human experts especially in spatial orientation, a fundamental SI factor. Moreover, we demonstrate a positive correlation between a model's spatial reasoning proficiency and its performance on an embodied AI task.

We introduce Latent Action Pretraining for general Action models (LAPA), the first unsupervised method for pretraining Vision-Language-Action (VLA) models without ground-truth robot action labels. Existing Vision-Language-Action models require action labels typically collected by human teleoperators during pretraining, which significantly limits possible data sources and scale. In this work, we propose a method to learn from internet-scale videos that do not have robot action labels. We first train an action quantization model leveraging VQ-VAE-based objective to learn discrete latent actions between image frames, then pretrain a latent VLA model to predict these latent actions from observations and task descriptions, and finally finetune the VLA on small-scale robot manipulation data to map from latent to robot actions. Experimental results demonstrate that our method significantly outperforms existing techniques that train robot manipulation policies from large-scale videos. Furthermore, it outperforms the state-of-the-art VLA model trained with robotic action labels on real-world manipulation tasks that require language conditioning, generalization to unseen objects, and semantic generalization to unseen instructions. Training only on human manipulation videos also shows positive transfer, opening up the potential for leveraging web-scale data for robotics foundation models.

One of the principal challenges in building VLM-powered GUI agents is visual grounding—localizing the appropriate screen region for action execution based on both the visual content and the textual plans. Most existing work formulates this as a text-based coordinate generation task. However, these approaches suffer from several limitations: weak spatial-semantic alignment due to lack of explicit spatial supervision; inability to handle ambiguous supervision targets, as single-point predictions penalize valid variations; and a mismatch between the dense nature of screen coordinates and the coarse, patch-level granularity of visual features extracted by models like Vision Transformers. In this paper, we propose **GUI-Actor**, a VLM-based method for coordinate-free GUI grounding. At its core, **GUI-Actor** introduces an attention-based action head that learns to align a dedicated `<ACTOR>` token with all relevant visual patch tokens, enabling the model to propose one or more action regions in a single forward pass. In line with this, we further design a grounding verifier to evaluate and select the most plausible action region from the candidates proposed for action execution. Extensive experiments show that **GUI-Actor** outperforms prior state-of-the-art methods on multiple GUI action grounding benchmarks, with improved generalization to unseen screen resolutions and layouts. Notably, **GUI-Actor-7B** achieves scores of **40.7** with Qwen2-VL and **44.6** with Qwen2.5-VL as backbones, outperforming **UI-TARS-72B (38.1)** on ScreenSpot-Pro, with significantly fewer parameters and training data. Furthermore, by incorporating the verifier, we find that fine-tuning only the newly introduced action head (~100M parameters for 7B model) while keeping the VLM backbone frozen is sufficient to achieve performance comparable to previous state-of-the-art models, highlighting that **GUI-Actor** can endow the underlying VLM with effective grounding capabilities without compromising its general-purpose strengths. Project page: [https://aka.ms/GUI-Actor](https://aka.ms/GUI-Actor)

Spatial reasoning in 3D space is central to human cognition and indispensable for embodied tasks such as navigation and manipulation. However, state-of-the-art vision–language models (VLMs) struggle frequently with tasks as simple as anticipating how a scene will look after an egocentric motion: they perceive 2D images but lack an internal model of 3D dynamics. We therefore propose SpatialNavigator, a test-time scaling framework that grants a VLM with this missing capability by coupling it to a controllable world model based on video diffusion. The VLM iteratively sketches a concise camera trajectory, while the world model synthesizes the corresponding view at each step. The VLM then reasons over this multi-view evidence gathered during the interactive exploration. Without any fine-tuning, our SpatialNavigator achieves an average 7.7\% performance boost on the representative spatial reasoning benchmark SAT, showing that pairing VLMs with world models for test-time scaling offers a simple, plug-and-play route to robust 3D reasoning. Meanwhile, our method also improves upon the test-time inference VLMs trained through reinforcement learning, which demonstrates the potential of our method that utilizes world models for test-time scaling.

Long video question answering is a challenging task that involves recognizing short-term activities and reasoning about their fine-grained relationships. State-of-the-art video Large Language Models (vLLMs) hold promise as a viable solution due to their demonstrated emergent capabilities on new tasks. However, despite being trained on millions of short seconds-long videos, vLLMs are unable to understand minutes-long videos and accurately answer questions about them. To address this limitation, we propose a lightweight and self-supervised approach, Key frame-conditioned long video-LLM (Koala), that introduces learnable spatiotemporal queries to adapt pretrained vLLMs for generalizing to longer videos. Our approach introduces two new tokenizers that condition on visual tokens computed from sparse video key frames for understanding short and long video moments. We train our proposed approach on HowTo100M and demonstrate its effectiveness on zero-shot long video understanding benchmarks, where it outperforms state-of-the-art large models by 3 - 6% in absolute accuracy across all tasks. Surprisingly, we also empirically show that our approach not only helps a pretrained vLLM to understand long videos but also improves its accuracy on short-term action recognition.

We propose a self-supervised approach for learning to perform audio source separation in videos based on natu-ral language queries, using only unlabeled video and au-dio pairs as training data. A key challenge in this task is learning to associate the linguistic description of a sound-emitting object to its visual features and the corresponding components of the audio waveform, all without access to annotations during training. To overcome this challenge, we adapt off-the-shelf vision-language foundation models to provide pseudo-target supervision via two novel loss functions and encourage a stronger alignment between the audio, visual and natural language modalities. During inference, our approach can separate sounds given text, video and audio input, or given text and audio input alone. We demonstrate the effectiveness of our self-supervised approach on three audio-visual separation datasets, including MUSIC, SOLOS and AudioSet, where we outperform state-of-the-art strongly supervised approaches despite not using object detectors or text labels during training. Our project page including publicly available code can be found at https://cs-people.bu.edu/rxtan/projectsNAST.

We introduce the task of spatially localizing narrated interactions in videos. Key to our approach is the ability to learn to spatially localize interactions with self-supervision on a large corpus of videos with accompanying transcribed narrations. To achieve this goal, we propose a multilayer cross-modal attention network that enables effective optimization of a contrastive loss during training. We introduce a divided strategy that alternates between computing inter- and intra-modal attention across the visual and natural language modalities, which allows effective training via directly contrasting the two modalities' representations. We demonstrate the effectiveness of our approach by self-training on the HowTo100M instructional video dataset and evaluating on a newly collected dataset of localized described interactions in the YouCook2 dataset. We show that our approach outperforms alternative baselines, including shallow co-attention and full cross-modal attention. We also apply our approach to grounding phrases in images with weak supervision on Flickr30K and show that stacking multiple attention layers is effective and, when combined with a word-to-region loss, achieves state of the art on recall-at-one and pointing hand accuracies.

The goal of weakly-supervised video moment retrieval is to localize the video segment most relevant to a description without access to temporal annotations during training. Prior work uses co-attention mechanisms to understand relationships between the vision and language data, but they lack contextual information between video frames that can be useful to determine how well a segment relates to the query. To address this, we propose an efficient Latent Graph Co-Attention Network (LoGAN) that exploits fine-grained frame-by-word interactions to jointly reason about the correspondences between all possible pairs of frames, providing context cues absent in prior work. Experiments on the DiDeMo and Charades-STA datasets demonstrate the effectiveness of our approach, where we improve Recall@1 by 520% over prior weakly-supervised methods, even boasting an 11% gain over strongly-supervised methods on DiDeMo, while also using significantly fewer model parameters than other co-attention mechanisms.

Large-scale dissemination of disinformation online intended to mislead or deceive the general population is a major societal problem.Rapid progression in image, video, and natural language generative models has only exacerbated this situation and intensified our need for an effective defense mechanism.While existing approaches have been proposed to defend against neural fake news, they are generally constrained to the very limited setting where articles only have text and metadata such as the title and authors.In this paper, we introduce the more realistic and challenging task of defending against machine-generated news that also includes images and captions.To identify the possible weaknesses that adversaries can exploit, we create a NeuralNews dataset composed of 4 different types of generated articles as well as conduct a series of human user study experiments based on this dataset.In addition to the valuable insights gleaned from our user study, we provide a relatively effective approach based on detecting visualsemantic inconsistencies, which will serve as an effective first line of defense and a useful reference for future work in defending against machine-generated disinformation.Our code and dataset can be downloaded from here.

Shouldn't language and vision features be treated equally in vision-language (VL) tasks? Many VL approaches treat the language component as an afterthought, using simple language models that are either built upon fixed word embeddings trained on text-only data or are learned from scratch. We conclude that language features deserve more attention, which has been informed by experiments which compare different word embeddings, language models, and embedding augmentation steps on five common VL tasks: image-sentence retrieval, image captioning, visual question answering, phrase grounding, and text-to-clip retrieval. Our experiments provide some striking results; an average embedding language model outperforms a LSTM on retrieval-style tasks; state-of-the-art representations such as BERT perform relatively poorly on vision-language tasks. From this comprehensive set of experiments we can propose a set of best practices for incorporating the language component of vision-language tasks. To further elevate language features, we also show that knowledge in vision-language problems can be transferred across tasks to gain performance with multi-task training. This multi-task training is applied to a new Graph Oriented Vision-Language Embedding (GrOVLE), which we adapt from Word2Vec using WordNet and an original visual-language graph built from Visual Genome, providing a ready-to-use vision-language embedding: http://ai.bu.edu/grovle.

Many real-world tasks require models to compare images along multiple similarity conditions (e.g. similarity in color, category or shape). Existing methods often reason about these complex similarity relationships by learning condition-aware embeddings. While such embeddings aid models in learning different notions of similarity, they also limit their capability to generalize to unseen categories since they require explicit labels at test time. To address this deficiency, we propose an approach that jointly learns representations for the different similarity conditions and their contributions as a latent variable without explicit supervision. Comprehensive experiments across three datasets, Polyvore-Outfits, Maryland-Polyvore and UT-Zappos50k, demonstrate the effectiveness of our approach: our model outperforms the state-of-the-art methods, even those that are strongly supervised with pre-defined similarity conditions, on fill-in-the-blank, outfit compatibility prediction and triplet prediction tasks. Finally, we show that our model learns different visually-relevant semantic sub-spaces that allow it to generalize well to unseen categories.