Browse 98 reviewed models and checkpoints connected to computer-use, GUI, web, and mobile agents. Each entry links to primary evidence and records its scope.
Review means AI-assisted editorial inspection, not independent reproduction. Unknown licenses and specifications remain unknown.
8B on-device smartphone agent built on MiniCPM-V that operates Chinese and English apps from screenshots, trained with reinforcement fine-tuning and a compact JSON action space.
Release date not recorded · androidAn eight-billion-parameter MiniCPM-V-based model that consumes smartphone screenshots and predicts compact actions for Chinese and English Android app tasks.
Release date not recorded · androidVision-based GUI agent models with unified action representations.
Release date not recorded · cross-platformMixture-of-experts GUI grounding model with 3.9B activated parameters that can use text or interleaved image history, with released training data and an AndroidWorld agent.
Release date not recorded · cross-platformThe context-aware Aria-UI release that uses interaction context to support grounding in dynamic Android and desktop agent tasks.
Release date not recorded · android, desktopUI-TARS-1.5-7B-derived checkpoint trained with ARPO on OSWorld desktop tasks.
Release date not recorded · desktop, linuxAutoGLM phone-agent checkpoint for Open-AutoGLM, fine-tuned from GLM-4.1V-9B-Base with about 10.3B BF16 parameters and tagged for Chinese.
Release date not recorded · android, iosMultilingual variant of the AutoGLM phone-agent checkpoint for Open-AutoGLM, likewise fine-tuned from GLM-4.1V-9B-Base with about 10.3B BF16 parameters.
Release date not recorded · android, iosMobile GUI agent research release with technical report, benchmark data, and evaluation tooling from vivo AI Lab.
Release date not recorded · androidClaude tool capability for screenshot-based computer interaction.
Release date not recorded · cross-platformReleased GUI world-model checkpoint that turns a screenshot and proposed action into HTML for rendering a predicted next screen.
Release date not recorded · androidVisual language model for GUI understanding and interaction.
Release date not recorded · cross-platformA bilingual GLM-4V-based execution model that uses screenshots, task instructions and action history to predict GUI operations.
Release date not recorded · desktop, windows, macos, cross-platformGUI grounding research using modality-aware visual reasoning and test-time scaling.
Release date not recorded · cross-platformWebDreamer model checkpoint that predicts the consequences of web actions from screenshots and textual descriptions.
Release date not recorded · webComputer-use model and training resources from the EvoCUA project.
Release date not recorded · desktopLarger EvoCUA computer-use agent checkpoint (20260105 release) with about 33B BF16 parameters in the Qwen3-VL architecture, tagged for OSWorld and for English and Chinese.
Release date not recorded · desktopEvoCUA computer-use agent checkpoint (20260105 release) with about 8.8B BF16 parameters in the Qwen3-VL architecture, tagged for OSWorld and for English and Chinese.
Release date not recorded · desktopMicrosoft browser-use agent models that act from screenshots by predicting click coordinates without accessibility trees; Fara1.5 comes in 4B, 9B and 27B, succeeding Fara-7B.
Release date not recorded · webMicrosoft's Fara checkpoint, a Qwen2.5-VL-architecture image-text-to-text model with about 8.3B BF16 parameters, English, MIT-licensed and linked to arXiv paper 2511.19663.
Release date not recorded · webScreen understanding research within the Ferret model family.
Release date not recorded · android, iosOpen mobile GUI agent stack pairing a 4B model that runs on consumer hardware with local multi-device infrastructure, an MCP server, and AndroidDaily benchmark tasks.
Release date not recorded · androidPreview GELab-Zero phone-use agent checkpoint fine-tuned from Qwen3-VL-4B-Instruct, with about 4.4B BF16 parameters, tagged for Android in English and Chinese, under Apache-2.0.
Release date not recorded · androidGemini capability for generating interface actions from screenshots.
Release date not recorded · cross-platformA Qwen-2.5-7B-Instruct checkpoint fine-tuned on Go-Browse WebArena trajectories for website task execution.
Release date not recorded · webScreenshot grounding checkpoint that emits a single click in a normalized coordinate space.
Release date not recorded · desktop, web, cross-platformGUI grounding model with attention-based action localization.
Release date not recorded · cross-platformGUI-Actor checkpoint fine-tuned from Qwen2-VL-7B-Instruct, with about 8.4B BF16 parameters, whose config names a pointer variant of the Qwen2-VL class rather than the stock one, under MIT.
Release date not recorded · cross-platformGUI-grounding training code that modifies reasoning templates, reward functions, and the GRPO objective, with a linked model release.
Release date not recorded · cross-platformA Qwen2.5-VL-3B checkpoint released for R1-style visual grounding experiments, predicting interface target locations from screenshots and instructions.
Release date not recorded · cross-platformGUI grounding models at 3B and 7B trained with Gaussian rewards for click proximity to element centers and region coverage, with variance scaled to element size.
Release date not recorded · cross-platformA Qwen2.5-VL grounding checkpoint trained with Gaussian point and coverage rewards whose granularity adapts to interface-element scale.
Release date not recorded · cross-platformA Qwen2.5-VL-based GUI checkpoint released with Mobile-Agent-v3 for grounded actions and specialized roles in its agent framework.
Release date not recorded · android, desktopGUI action-model training and evaluation code with verifiable rewards, plus linked reinforcement-learning data and checkpoints.
Release date not recorded · cross-platformA single Hub release containing GUI-R1 3B and 7B action-model checkpoints trained with reinforcement learning across interface platforms.
Release date not recorded · desktop, windows, linux, macos, android, webH computer-use model family and its release-specific access options.
Release date not recorded · cross-platformA Qwen2.5-VL-based action model released for the Surfer-H web-agent system, supporting policy, interface-localization and validation roles.
Release date not recorded · webA Qwen3-VL-based checkpoint for navigation and multi-step GUI actions across web, desktop and mobile interfaces.
Release date not recorded · web, desktop, cross-platformA sparse Qwen3.5-based GUI-action checkpoint released for navigation and computer-use workflows.
Release date not recorded · web, desktop, cross-platformA Qwen3.6-based computer-use checkpoint designed to request GUI actions, code execution and tool calls through an external agent harness.
Release date not recorded · desktop, webInfiGUI-G1 3B and 7B grounding models trained with Adaptive Exploration Policy Optimization, which samples several candidate answers per pass under an adaptive exploration reward.
Release date not recorded · cross-platformA GUI-grounding checkpoint trained with adaptive exploration rewards and multiple candidate answers to improve instruction-to-element alignment.
Release date not recorded · cross-platform3B GUI agent trained in two stages: distilling spatial reasoning from teacher models, then reinforcement learning that rewards accurate sub-goals and trains error recovery.
Release date not recorded · cross-platformA Qwen2.5-VL checkpoint trained with the Actor2Reasoner procedure to combine interface grounding with planning and recovery reasoning.
Release date not recorded · web, desktop, androidA Llama 3 8B instruction-model checkpoint fine-tuned on WebLINX web interactions and dialogue, predicting browser actions from a textual state representation.
Release date not recorded · webMicrosoft foundation model for multimodal agents, pretrained with Set-of-Mark and Trace-of-Mark objectives on UI, robotics and video data and fine-tunable for web and mobile UI navigation.
Release date not recorded · webA released MAI-UI checkpoint combining mobile GUI actions, agent-user interaction and tool-use-oriented navigation.
Release date not recorded · androidA GLM-4.1V-based mobile action checkpoint released by the MobileRL project for Android interaction.
Release date not recorded · androidA screen-parsing weight bundle containing icon-region detectors and an icon captioner for converting screenshots into agent-readable interface elements.
Release date not recorded · unspecifiedComputer-using model introduced with the Operator research preview.
Release date not recorded · web, desktopOpen computer-use research stack including models and training resources.
Release date not recorded · desktopA Qwen2.5-VL-based computer-use checkpoint that combines screenshot history with planning and grounded executable actions.
Release date not recorded · desktop, linux, windows, macosA Qwen2.5-VL-based computer-use checkpoint that combines screenshot history and reflective reasoning to predict executable desktop actions.
Release date not recorded · desktop, linux, windows, macosMobile-agent checkpoint based on Qwen3.6-27B for GUI and app-native tool interaction.
Release date not recorded · android, webMobile-agent checkpoint trained on OpenMobile-Data-v2 for screenshot actions and app-native tool calls.
Release date not recorded · android, webOS-Atlas GUI action models, including grounding bases fine-tuned from InternVL2-4B and Qwen2-VL-7B-Instruct that accept any image size and output 0-1000 relative coordinates.
Release date not recorded · cross-platformOS-Atlas base checkpoint with about 4.1B parameters, fine-tuned from InternVL2-4B and loaded in Transformers through custom InternVL chat model code.
Release date not recorded · cross-platformOS-Atlas base checkpoint fine-tuned from Qwen2-VL-7B-Instruct, using the standard Qwen2-VL architecture in Transformers with about 8.3B BF16 parameters.
Release date not recorded · cross-platformThe AndroidControl-oriented Qwen2-VL checkpoint trained with mobile trajectories generated through OS-Genesis reverse task synthesis.
Release date not recorded · androidA released LLaVA checkpoint for visual browser interaction from the PAE autonomous task-proposal and policy-learning system.
Release date not recorded · webComputer-use checkpoint fine-tuned on PC-Agent-E trajectories; the model-card prose identifies Qwen2.5-VL-72B-Instruct as its starting model.
Release date not recorded · desktop, windows, linuxMicrosoft GUI grounding models from an empirical study of grounding training, from data collection to model training, released as Phi-Ground-4B-7C with evaluation code.
Release date not recorded · cross-platformMicrosoft Phi-Ground GUI grounding checkpoint fine-tuned from Phi-3.5-vision-instruct, shipped as two PyTorch weight shards with custom Phi-3-V configuration and processing code under MIT.
Release date not recorded · cross-platformThe planner component of Plan-and-Act, generating structured high-level plans that an executor turns into web actions.
Release date not recorded · webPOINTS-GUI-G-8B GUI grounding model, trained from a POINTS-1.5 base without native grounding through a data pipeline that unifies coordinates, filters noise and adds hard cases.
Release date not recorded · cross-platformA specialized interface-grounding checkpoint that returns normalized point coordinates for a requested element in a screenshot.
Release date not recorded · web, desktop, cross-platformComputer-use model family trained with cross-platform interaction data.
Release date not recorded · cross-platformA Qwen2.5-VL-based computer-use checkpoint with direct grounding and agent action modes across desktop, mobile and web interfaces.
Release date not recorded · web, desktop, android, linuxGUI grounding model trained to locate visual interface elements.
Release date not recorded · cross-platformThe released SeeClick visual grounding checkpoint for mapping an interface screenshot and referring instruction to a point or bounding box.
Release date not recorded · web, desktop, android, iosSeg2Click checkpoint based on Qwen2.5-VL-3B with a custom action-mask decoder.
Release date not recorded · cross-platformCompact Seg2Click checkpoint with a merged Qwen3.5 backbone and an action-mask decoder.
Release date not recorded · cross-platformSeg2Click checkpoint using a 2B Qwen3.5 backbone to predict a mask and select an interface click.
Release date not recorded · cross-platformSeg2Click checkpoint pairing a 4B Qwen3.5 backbone with pixel-level action-mask prediction.
Release date not recorded · cross-platformSeg2Click checkpoint pairing a 9B Qwen3.5 backbone with a decoder for valid GUI click regions.
Release date not recorded · cross-platformLightweight GUI-agent vision-language-action model from Show Lab and Microsoft, released with UI-guided token selection, training code, the ShowUI-web and ShowUI-desktop datasets and vLLM inference.
Release date not recorded · cross-platformShowUI vision-language-action checkpoint fine-tuned from Qwen2-VL-2B-Instruct, published as a single PyTorch weight file under the MIT license and tagged for GUI agents and computer use.
Release date not recorded · cross-platformVision-language-action model and training code for GUI dragging, with links to checkpoints and continuous-action evaluation data.
Release date not recorded · desktopA compact visual-action checkpoint that predicts continuous GUI drag trajectories from a screenshot and a natural-language instruction.
Release date not recorded · unspecifiedAn InternVL2-based visual GUI checkpoint distributed for grounding and interface navigation, with the publisher recommending task-specific fine-tuning of the base release.
Release date not recorded · cross-platformCode, training data and weights for universal visual grounding of GUI elements from screenshots, including Qwen2-VL-based UGround-V1 at 2B, 7B and 72B and offline and online evaluation suites.
Release date not recorded · cross-platformUGround-V1 checkpoint fine-tuned from Qwen2-VL-7B, with about 8.3B BF16 parameters in the standard Qwen2-VL architecture, provided as safetensors and a PyTorch file under Apache-2.0.
Release date not recorded · cross-platformGUI agent trained by RL with a continuous reward, a simple-thinking reward and cropping-based resampling on about 9k samples, plus a plug-in decomposed grounding-and-selection inference method.
Release date not recorded · cross-platformThe released UI-AGILE model for GUI grounding with reward shaping, crop-based resampling and decomposed selection during inference.
Release date not recorded · androidA Qwen2.5-VL mobile action model trained using existing and reward-guided synthetic trajectories from the UI-Genie framework.
Release date not recorded · androidRule-based reinforcement learning for GUI action prediction, releasing UI-R1-3B, trained from Qwen2.5-VL-3B, the efficiency-oriented UI-R1-E-3B, and training data.
Release date not recorded · cross-platformA Qwen2.5-VL checkpoint trained with reinforcement learning for GUI action prediction and evaluated on mobile, desktop and web grounding tasks.
Release date not recorded · web, desktop, androidA mobile GUI checkpoint trained with semi-online reinforcement learning that uses offline trajectories to improve multi-turn action prediction.
Release date not recorded · androidNative GUI agent models; UI-TARS-1.5 adds reinforcement-learned reasoning before acting and an open 7B checkpoint, and UI-TARS-2 extends to games, code and tool use.
Release date not recorded · cross-platformUI-TARS 1.5 GUI-agent checkpoint in the Qwen2.5-VL architecture with about 8.3B parameters stored in FP32, released under Apache-2.0 for English image-text-to-text use in Transformers.
Release date not recorded · cross-platformGUI agent model family; UI-Venus-2, at 9B and 27B, acts on mobile, web and desktop in one end-to-end agent, trained with keypoint-based multi-judge task verification.
Release date not recorded · cross-platformUI-Venus 2 GUI-agent checkpoint fine-tuned from Qwen3.5-9B, with about 9.4B BF16 parameters in the Qwen3.5 multimodal architecture, loaded through Transformers for image-text-to-text.
Release date not recorded · cross-platformA MultiUI-trained visual-language checkpoint for text-rich webpage understanding and interface tasks.
Release date not recorded · webA visual trajectory judge that retains selected intermediate screenshots to evaluate whether a web agent completed its task.
Release date not recorded · webA WebRL browser-policy checkpoint for performing actions in the five website environments of WebArena.
Release date not recorded · webThe outcome-supervision checkpoint used by WebRL to score browser trajectories during policy training.
Release date not recorded · webA public text-based process-reward checkpoint from the Web-Shepherd collection for scoring web-agent decisions against task checklists.
Release date not recorded · webQwen2.5-VL-3B-based GUI grounding checkpoint that predicts click locations from screenshots and text instructions across desktop, mobile and web screens.
Release date not recorded · cross-platform, desktop, web