Alternatives to HunyuanWorld

Most AI video tools hand you a black box: closed weights, a subscription, and no way to see what is happening under the hood. LTX takes the opposite approach. Built by Lightricks, LTX is an open foundation model that generates and simulates across video, audio, and the physical world, and it puts the weights, the code, and the control in your hands. At the center of the model is LTX-2.5, a 22B-parameter dual-stream diffusion transformer that produces native 4K video at up to 50 frames per second, with audio and video generated together in a single pass rather than stitched together afterward. Artificial Analysis, an independent benchmarking group, currently ranks LTX among the top three AI video models in the world. You choose how you want to use it. Download the open weights and run LTX-2.5 on your own hardware. License the model for on-premise deployment backed by enterprise support. Or build directly on LTX Studio, the production suite that turns the model into a full creative workflow. Companies like ElevenLabs, Asteria Film Co., Magnopus, and NVIDIA already rely on LTX for their own work. LTX is not built for one-off social clips. It is infrastructure for teams that generate motion, audio, and physical environments as part of their own products and pipelines.

Tencent has unveiled the Hunyuan T1, its advanced AI model, which is now accessible to all users via the Tencent Yuanbao platform. This model is particularly adept at grasping various dimensions and potential logical connections, making it ideal for tackling intricate challenges. Users have the opportunity to explore a range of AI models available on the platform, including DeepSeek-R1 and Tencent Hunyuan Turbo. Anticipation is building for the forthcoming official version of the Tencent Hunyuan T1 model, which will introduce external API access and additional services. Designed on the foundation of Tencent's Hunyuan large language model, Yuanbao stands out for its proficiency in Chinese language comprehension, logical reasoning, and effective task performance. It enhances user experience by providing AI-driven search, summaries, and writing tools, allowing for in-depth document analysis as well as engaging prompt-based dialogues. The platform's versatility is expected to attract a wide array of users seeking innovative solutions.

WorldClaw represents a comprehensive framework that facilitates the generation of expansive, freely navigable, and modifiable 3D open worlds, all derived from open-ended text prompts. Instead of constructing an entire world in one go, it employs a strategy that transitions from a broad overview to detailed regional features, ensuring spatial coherence while enriching local attributes. Initially, planning agents convert the text prompt into a structured scene specification that includes regions, terrain, assets, materials, visual aesthetics, and spatial arrangements. It establishes a globally consistent terrain base by utilizing semantic layouts, reusable assets, generative or procedural materials, and height fields that are sensitive to regional context. For areas that demand more intricate detail, WorldClaw generates terrain-conditioned compositions, reconstructs editable textured meshes, and appropriately places them within the scene. Subsequently, render-based agents enhance the terrain geometry, fine-tune object appearances, optimize arrangements, and ensure proper interactions with the surrounding environment. This multi-layered approach allows for both the creation of vast landscapes and the intricate detailing needed for immersive exploration.

HunyuanVideo is a cutting-edge video generation model powered by AI, created by Tencent, that expertly merges virtual and real components, unlocking endless creative opportunities. This innovative tool produces videos of cinematic quality, showcasing smooth movements and accurate expressions while transitioning effortlessly between lifelike and virtual aesthetics. By surpassing the limitations of brief dynamic visuals, it offers complete, fluid actions alongside comprehensive semantic content. As a result, this technology is exceptionally suited for use in various sectors, including advertising, film production, and other commercial ventures, where high-quality video content is essential. Its versatility also opens doors for new storytelling methods and enhances viewer engagement.

Tencent's Hunyuan-TurboS represents a cutting-edge AI model crafted to deliver swift answers and exceptional capabilities across multiple fields, including knowledge acquisition, mathematical reasoning, and creative endeavors. Departing from earlier models that relied on "slow thinking," this innovative system significantly boosts response rates, achieving a twofold increase in word output speed and cutting down first-word latency by 44%. With its state-of-the-art architecture, Hunyuan-TurboS not only enhances performance but also reduces deployment expenses. The model skillfully integrates fast thinking—prompt, intuition-driven responses—with slow thinking—methodical logical analysis—ensuring timely and precise solutions in a wide array of situations. Its remarkable abilities are showcased in various benchmarks, positioning it competitively alongside other top AI models such as GPT-4 and DeepSeek V3, thus marking a significant advancement in AI performance. As a result, Hunyuan-TurboS is poised to redefine expectations in the realm of artificial intelligence applications.

Tencent Hunyuan represents a comprehensive family of multimodal AI models crafted by Tencent, encompassing a range of modalities including text, images, video, and 3D data, all aimed at facilitating general-purpose AI applications such as content creation, visual reasoning, and automating business processes. This model family features various iterations tailored for tasks like natural language interpretation, multimodal comprehension that combines vision and language (such as understanding images and videos), generating images from text, creating videos, and producing 3D content. The Hunyuan models utilize a mixture-of-experts framework alongside innovative strategies, including hybrid "mamba-transformer" architectures, to excel in tasks requiring reasoning, long-context comprehension, cross-modal interactions, and efficient inference capabilities. A notable example is the Hunyuan-Vision-1.5 vision-language model, which facilitates "thinking-on-image," allowing for intricate multimodal understanding and reasoning across images, video segments, diagrams, or spatial information. This robust architecture positions Hunyuan as a versatile tool in the rapidly evolving field of AI, capable of addressing a diverse array of challenges.

Atlas represents an advanced omni world model designed for spatial intelligence, seamlessly engaging with text, images, video, and 3D data. As a sophisticated multimodal autoregressive diffusion transformer, it integrates various inputs into a common spatial framework while predicting subsequent elements and ensuring coherence in 3D with its observations and creative visions. This powerful model facilitates world generation, reconstruction, and simulation across diverse applications. It possesses the capability to produce images and videos based on one or multiple reference images, offering precise camera control to yield lengthy and coherent videos following custom-designed camera trajectories. In terms of spatial reconstruction, Atlas excels at reimagining real-world environments from limited input images, allowing it to generate unique perspectives and deliver explicit 3D outputs like point clouds and 3D Gaussian splats. With the addition of more input views, the model can further enhance context, leading to increasingly accurate reconstructions with less reliance on imagination. Furthermore, Atlas also analyzes and predicts the evolution of worlds over time, adding a dynamic layer to its capabilities.

Pika Soundtrack is an innovative model that transforms silent videos into rich audio experiences by integrating motion-sensitive sound effects, music, ambient noises, and voiceovers that align perfectly with the visual content. Users have the option to leave the input prompt empty for the model to create a complete soundscape automatically or to provide specific instructions regarding which elements to highlight, include, or exclude. Unlike conventional methods that merely attach sounds to videos, this model comprehensively analyzes the scene, ensuring that every sound is precisely timed and that all audio components remain consistent throughout the video. This thoughtful synchronization allows for a seamless blend of sound effects, ambient sounds, music, and dialogue, giving the impression that they all naturally coexist within the same environment. According to Pika's testing, Soundtrack outperformed other models like LTX-2.3 Foley V2A, HunyuanVideo-Foley, and MMAudio v2 in achieving the best semantic coherence and minimal audiovisual misalignment in its full-duration benchmark. The ability to capture the essence of a scene while maintaining audio clarity makes Pika Soundtrack a standout choice for video creators looking to enhance their content.

HunyuanVision, an innovative vision-language model created by Tencent's Hunyuan team, employs a mamba-transformer hybrid architecture that excels in performance and offers efficient inference for multimodal reasoning challenges. The latest iteration, Hunyuan-Vision-1.5, focuses on the concept of “thinking on images,” enabling it to not only comprehend the interplay of visual and linguistic content but also engage in advanced reasoning that includes tasks like cropping, zooming, pointing, box drawing, or annotating images for enhanced understanding. This model is versatile, supporting various vision tasks such as image and video recognition, OCR, and diagram interpretation, in addition to facilitating visual reasoning and 3D spatial awareness, all within a cohesive multilingual framework. Designed for compatibility across different languages and tasks, HunyuanVision aims to be open-sourced, providing access to checkpoints, a technical report, and inference support to foster community engagement and experimentation. Ultimately, this initiative encourages researchers and developers to explore and leverage the model's capabilities in diverse applications.

Happy Oyster is a dynamic AI platform that serves as a world model, enabling users to create, investigate, and continually refine immersive 3D environments using straightforward prompts. Rather than generating a static result, it functions as a responsive ecosystem that adapts in real time to user interactions, allowing for updates to scenes based on commands delivered through text, voice, or visual inputs. The platform promotes multimodal engagement and upholds consistent physical principles such as lighting, gravity, and motion, ensuring that the environments act like coherent, enduring worlds instead of fragmented scenes. It features two primary modes: Directing, where users have the power to steer scenes, modify camera perspectives, control characters, and influence unfolding narratives; and Wandering, which allows users to delve into an infinitely expansive world from a first-person viewpoint, freely navigating beyond the initial frames. This dual functionality enhances user experience by providing both creative control and exploratory freedom.

HunyuanCustom is an advanced framework for generating customized videos across multiple modalities, focusing on maintaining subject consistency while accommodating conditions related to images, audio, video, and text. This framework builds on HunyuanVideo and incorporates a text-image fusion module inspired by LLaVA to improve multi-modal comprehension, as well as an image ID enhancement module that utilizes temporal concatenation to strengthen identity features throughout frames. Additionally, it introduces specific condition injection mechanisms tailored for audio and video generation, along with an AudioNet module that achieves hierarchical alignment through spatial cross-attention, complemented by a video-driven injection module that merges latent-compressed conditional video via a patchify-based feature-alignment network. Comprehensive tests conducted in both single- and multi-subject scenarios reveal that HunyuanCustom significantly surpasses leading open and closed-source methodologies when it comes to ID consistency, realism, and the alignment between text and video, showcasing its robust capabilities. This innovative approach marks a significant advancement in the field of video generation, potentially paving the way for more refined multimedia applications in the future.

FLUX 3 Action is a versatile 7B world-action model aimed at enhancing action prediction for robotics and other environments with high latency demands. This model is built upon the multimodal FLUX 3 backbone and has undergone extensive pretraining on vast datasets encompassing images, videos, and audio, with a notable focus on video content. It then goes through a phase of joint video-action training and fine-tuning tailored to specific robotic applications and action frameworks. By utilizing instructions, visual input from cameras, and robot joint angles, FLUX 3 Action is capable of predicting motor commands alongside anticipated visual outcomes, enabling robots to perform actions, reassess their surroundings, and iterate on their plans. This approach contrasts with methods that treat visual prediction and control as separate processes, as FLUX 3 Action integrates future video predictions with action commands, effectively leveraging knowledge gained from extensive video pretraining to refine robot control mechanisms. Impressively, its single-step 7B model achieves a success rate of 38.3% on the RoboLab-120 benchmark, showcasing its effectiveness in real-world applications. Furthermore, this innovative integration of action and perception marks a significant advancement in the field of robotic control.

SKY-001 is an innovative engineering world model introduced by Skyfall AI, designed to create structured visual design plans based on user prompts, reference images, and spatial parameters. It disassembles engineered entities into distinct assemblies and components, subsequently forecasting the sequences for sketches and 3D construction states related to each part. The capabilities of this model are showcased through computer-aided design (CAD) tasks, including the reconstruction of mechanical assemblies, as well as in building information modeling (BIM) tasks, which involve translating building outlines and specifications into detailed floor plans. Its hierarchical approach allows for parallel exploration of various design candidates, enhances the refinement of component geometries, and facilitates planning across numerous construction phases. Skyfall AI promotes SKY-001 as a research tool for applications in design and engineering, while also hinting at possible expansions into other areas of physical design. Those interested in utilizing this innovative tool can find a link to join a waitlist for future access on the product page, ensuring they remain updated on its public release. This model represents a significant step forward in the integration of artificial intelligence into practical design workflows.

Genie 3 represents DeepMind's innovative leap in general-purpose world modeling, capable of real-time generation of immersive 3D environments at 720p resolution and 24 frames per second, maintaining consistency for several minutes. When provided with textual prompts, this advanced system fabricates interactive virtual landscapes that allow users and embodied agents to explore and engage with natural occurrences from various viewpoints, including first-person and isometric perspectives. One of its remarkable capabilities is the emergent long-horizon visual memory, which ensures that environmental details remain consistent even over lengthy interactions, retaining off-screen elements and spatial coherence when revisited. Additionally, Genie 3 features “promptable world events,” granting users the ability to dynamically alter scenes, such as modifying weather conditions or adding new objects as desired. Tailored for research involving embodied agents, Genie 3 works in harmony with systems like SIMA, enhancing navigation based on specific goals and enabling the execution of intricate tasks. This level of interactivity and adaptability marks a significant advancement in how virtual environments can be experienced and manipulated.

Text2Mesh generates intricate geometric and color details across various source meshes, guided by a specified text prompt. The results of our stylization process seamlessly integrate unique and seemingly unrelated text combinations, effectively capturing both overarching semantics and specific part-aware features. Our system, Text2Mesh, enhances a 3D mesh by predicting colors and local geometric intricacies that align with the desired text prompt. We adopt a disentangled representation of a 3D object, using a fixed mesh as content integrated with a learned neural network, which we refer to as the neural style field network. To alter the style, we compute a similarity score between the style-describing text prompt and the stylized mesh by leveraging CLIP's representational capabilities. What sets Text2Mesh apart is its independence from a pre-existing generative model or a specialized dataset of 3D meshes. Furthermore, it is capable of processing low-quality meshes, including those with non-manifold structures and arbitrary genus, without the need for UV parameterization, thus enhancing its versatility in various applications. This flexibility makes Text2Mesh a powerful tool for artists and developers looking to create stylized 3D models effortlessly.

Reactor is currently developing an essential layer for world models and is inviting users to engage with real-time world models in an early preview. The core of its product strategy revolves around worlds that are generated on the spot, allowing for instantaneous creation of pixels, sounds, and actions, which transforms user interaction with both software and the tangible world. This preview marks the beginning of a new era, enabling users to explore AI-generated environments powered by a global low-latency infrastructure. Reactor is dedicated to pioneering the next wave of AI, focusing on real-time world models that can be navigated by people, agents, and robots in a frame-by-frame manner. Instead of merely presenting generated video as a passive viewing experience, Reactor envisions interactive spaces that can be lived in, manipulated, and molded as they unfold. The research and product development prioritize real-time interactions, inference, customizable world models, and systems capable of making dynamic visual settings responsive enough for live engagement, paving the way for a more immersive experience. This innovative approach aims to redefine the boundaries of digital interaction, merging creativity with cutting-edge technology.

Mirage 2 is an innovative Generative World Engine powered by AI, allowing users to effortlessly convert images or textual descriptions into dynamic, interactive game environments right within their browser. Whether you upload sketches, concept art, photographs, or prompts like “Ghibli-style village” or “Paris street scene,” Mirage 2 crafts rich, immersive worlds for you to explore in real time. This interactive experience is not bound by pre-defined scripts; users can alter their environments during gameplay through natural-language chat, enabling the settings to shift fluidly from a cyberpunk metropolis to a lush rainforest or a majestic mountaintop castle, all while maintaining low latency (approximately 200 ms) on a standard consumer GPU. Furthermore, Mirage 2 boasts smooth rendering and offers real-time prompt control, allowing for extended gameplay durations that go beyond ten minutes. Unlike previous world-modeling systems, it excels in general-domain generation, eliminating restrictions on styles or genres, and provides seamless world adaptation alongside sharing capabilities, which enhances collaborative creativity among users. This transformative platform not only redefines game development but also encourages a vibrant community of creators to engage and explore together.

Tencent HY is a versatile and comprehensive family of large models developed in-house by Tencent, designed to deliver AI solutions tailored for enterprise needs across various domains such as content creation, business automation, and real-world agent functionalities. It encompasses multiple modalities, including language, imagery, 3D modeling, and translation, seamlessly integrating Tencent’s proprietary algorithms with advanced natural language processing and computer vision technologies to enable superior image generation, 3D content creation, and intelligent applications. Via Tencent Hunyuan AI Studio, users can engage with the model through intuitive human-computer dialogue, which allows the system to comprehend commands, perform tasks, assist users in information retrieval, generate content, and discover the model's extensive capabilities in a user-friendly environment. Additionally, Tencent HY facilitates API integration and customizable parameter configurations, enhancing accessibility and usability for developers, product teams, and enterprise-focused applications. This adaptability ensures that a wide range of users can leverage the power of Tencent HY in their projects, driving innovation and efficiency across industries.

Tencent Hunyuan 3D is an innovative platform driven by artificial intelligence that focuses on the generation of 3D content. By utilizing cutting-edge AI technology, this platform enables users to efficiently produce lifelike and engaging 3D models and animations. Targeted primarily at sectors like gaming, virtual reality, and digital media, it provides a convenient solution for the creation of top-notch 3D assets. With its user-friendly interface, users can seamlessly bring their creative visions to life.

Odyssey-2 Pro represents a groundbreaking general-purpose world model that allows for the generation of continuous, interactive simulations, which can be seamlessly integrated into various products through the Odyssey API, akin to the significant impact that GPT-2 had on language processing. This model is developed using extensive video and interaction datasets, enabling it to understand the progression of events frame-by-frame and produce simulations that last for minutes, rather than just brief static clips. With its enhanced physics, richer dynamics, more lifelike behaviors, and clearer visuals, Odyssey-2 Pro streams 720p video at approximately 22 frames per second, providing immediate responses to user prompts and actions. Furthermore, it facilitates the integration of interactive streams, viewable streams, and parameterized simulations into applications through straightforward SDKs available in both JavaScript and Python. Developers can incorporate this powerful model with fewer than ten lines of code, allowing them to craft open-ended, interactive video experiences that dynamically change based on user interactions, thus enhancing the overall engagement and immersion. This capability not only revolutionizes how simulations are utilized but also opens the door for innovative applications across various industries.

Hy Image 3.5 represents the latest advancement from Tencent Hunyuan in the realm of image generation, aimed at enhancing the entire journey from understanding user intent to achieving visual representation. This model facilitates a cohesive workflow that encompasses text-to-image, image-to-image, reference-based generation, and multi-turn conversational editing. Users have the flexibility to input text along with reference images, maintain context across multiple interactions, and seamlessly refine or alter images without having to restart the creative journey. Moreover, it is capable of processing several reference images in one go, making it ideal for maintaining subject consistency, managing composition, creating product visuals, designing characters, producing advertising content, and engaging in iterative design processes. The model accommodates a variety of aspect ratios and output sizes, providing options for custom dimensions and high-resolution generation via the API. Additionally, the Hy Image 3.5 Preview utilizes a conversational message protocol, which allows users to articulate their image creation and editing commands in a natural and intuitive manner, thus enhancing the overall user experience. This innovative feature streamlines the creative process, making it more accessible and user-friendly.

Odyssey-2 Max is an advanced, real-time world simulation model that transcends conventional generative AI by learning the dynamics of the physical world and facilitating ongoing, interactive settings. As the third iteration in the Odyssey-2 series, it boasts a remarkable increase in scale, featuring three times more parameters and ten times the computational power compared to its predecessor, Odyssey-2 Pro, which fosters new emergent behaviors and enhances the stability and realism of simulations. Crafted to accurately replicate physics, human movement, interactions, and environmental changes in real time, it offers continuous visual output that adapts instantaneously to user commands rather than relying on fixed video clips. In contrast to traditional video models that produce short, predetermined sequences, Odyssey-2 Max enables the creation of extensive simulations that evolve in real time, allowing users to engage with a dynamically unfolding environment. This innovative approach redefines user interaction, making every session unique and immersive as the simulation adapts to each new input.

Agora-1 is an innovative multi-agent world model that facilitates real-time interaction among several participants, whether they are human or AI, within a shared world simulation. This model represents the inaugural installment in a sequence of multi-agent world models aimed at uncovering new shared experiences in various fields such as gaming, robotics, defense, education, and foundational models. Traditionally, world models have excelled at creating high-fidelity simulations of diverse environments but were limited by the fact that only one active participant could engage with these simulated worlds at a time. With Agora-1, the concept of multi-agent world simulations is brought to life, enabling as many as four players to engage simultaneously in the same generated environment. These players are immersed in a competitive deathmatch simulation, where each participant interacts with the same world concurrently, as the model adeptly simulates player actions, manages a unified world state, and streams the rendered visuals to every player, enhancing the immersive experience. This advancement paves the way for more collaborative and interactive engagements in various domains.

SAM 3D consists of a duo of sophisticated foundation models that can transform a typical RGB image into an impressive 3D representation of either objects or human figures. This system features SAM 3D Objects, which accurately reconstructs the complete 3D geometry, textures, and spatial arrangements of items found in real-world environments, effectively addressing challenges posed by clutter, occlusions, and varying lighting conditions. Additionally, SAM 3D Body generates dynamic human mesh models that capture intricate poses and shapes, utilizing the "Meta Momentum Human Rig" (MHR) format for enhanced detail. The design of this system allows it to operate effectively with images taken in natural settings without the need for further training or fine-tuning: users simply upload an image, select the desired object or individual, and receive a downloadable asset (such as .OBJ, .GLB, or MHR) that is instantly ready for integration into 3D software. Highlighting features like open-vocabulary reconstruction applicable to any object category, multi-view consistency, and occlusion reasoning, the models benefit from a substantial and diverse dataset containing over one million annotated images from the real world, which contributes significantly to their adaptability and reliability. Furthermore, the models are available as open-source, promoting wider accessibility and collaborative improvement within the development community.

Starchild-1 represents a groundbreaking advancement in real-time multimodal world modeling, designed to simultaneously replicate both visual and auditory experiences. In contrast to traditional language models that derive knowledge solely from text, world models like Starchild-1 learn from the actual environment through the analysis of pixels, movements, and actions captured in extensive video data, thereby gaining the ability to comprehend and simulate the evolving nature of the world. This innovative model surpasses previous world models, which typically concentrated only on visual output, by autoregressively generating coordinated audio and video in response to real-time user interactions. Rather than generating a static video segment, it forecasts the forthcoming audio and visual states of a scenario, influenced by historical data and real-time inputs, facilitating a dynamic interplay of environments, dialogues, background sounds, and world interactions. Users can actively contribute text, speech, and actions to the model as it operates, resulting in a continuously shifting auditory and visual landscape. This level of interactivity allows for a rich and immersive experience, reshaping how users engage with simulated environments.

GWM-1 is Runway’s first family of General World Models created to interact dynamically with simulated reality. Built on Gen-4.5, the model produces real-time, action-conditioned video rather than static imagery alone. GWM-1 allows users to control environments through camera motion, robotics commands, events, and speech inputs. It generates coherent visual scenes that persist across movement and time. The model supports synchronized video, image, and audio generation for immersive simulation. GWM-1 is designed to learn from interaction and trial-and-error rather than passive data consumption. It enables realistic exploration of both physical and imagined worlds. Runway positions GWM-1 as foundational technology for robotics, training, and creative systems. The model scales across multiple domains without manual environment design. GWM-1 marks a shift toward experiential AI systems.

AudioLM is an innovative audio language model designed to create high-quality, coherent speech and piano music by solely learning from raw audio data, eliminating the need for text transcripts or symbolic forms. It organizes audio in a hierarchical manner through two distinct types of discrete tokens: semantic tokens, which are derived from a self-supervised model to capture both phonetic and melodic structures along with broader context, and acoustic tokens, which come from a neural codec to maintain speaker characteristics and intricate waveform details. This model employs a series of three Transformer stages, initiating with the prediction of semantic tokens to establish the overarching structure, followed by the generation of coarse tokens, and culminating in the production of fine acoustic tokens for detailed audio synthesis. Consequently, AudioLM can take just a few seconds of input audio to generate seamless continuations that effectively preserve voice identity and prosody in speech, as well as melody, harmony, and rhythm in music. Remarkably, evaluations by humans indicate that the synthetic continuations produced are almost indistinguishable from actual recordings, demonstrating the technology's impressive authenticity and reliability. This advancement in audio generation underscores the potential for future applications in entertainment and communication, where realistic sound reproduction is paramount.

Odyssey-2 represents a cutting-edge interactive video technology that allows for immediate and real-time video generation that users can engage with. Simply enter a prompt, and the system promptly starts streaming several minutes of video that reacts to your input. This innovation transforms video from a traditional playback experience into a responsive, action-sensitive stream: the model operates in a causal and autoregressive manner, crafting each frame based on previous frames and your actions instead of adhering to a set timeline, which enables a seamless adaptation of camera perspectives, environments, characters, and narratives. The platform efficiently begins video streaming nearly instantaneously, generating new frames approximately every 50 milliseconds (around 20 frames per second), ensuring that you don’t have to wait long for content but instead immerse yourself in an evolving narrative. Beneath its surface, the model employs an advanced multi-stage training process that shifts from generating fixed clips to creating open-ended interactive video experiences, granting you the ability to type or voice commands while exploring a world crafted by AI that responds in real-time. This innovative approach not only enhances engagement but also revolutionizes the way viewers interact with visual storytelling.

Hunyuan Motion, often referred to as HY-Motion 1.0, represents an advanced AI model designed for transforming text into 3D motion, utilizing a billion-parameter Diffusion Transformer combined with flow matching techniques to create high-quality, skeleton-based animations in mere seconds. This innovative system comprehends detailed descriptions in both English and Chinese, allowing it to generate fluid and realistic motion sequences that can easily integrate into typical 3D animation workflows by exporting into formats like SMPL, SMPLH, FBX, or BVH, which are compatible with software such as Blender, Unity, Unreal Engine, and Maya. Its sophisticated training approach includes a three-phase pipeline: extensive pre-training on thousands of hours of motion data, meticulous fine-tuning on selected sequences, and reinforcement learning informed by human feedback, all of which significantly boost its capacity to interpret intricate commands and produce motion that is not only realistic but also temporally coherent. This model stands out for its ability to adapt to various animation styles and requirements, making it a versatile tool for creators in the gaming and film industries.

Odyssey-3 represents an advanced foundational world model and dynamic system that forecasts the movement and interaction of objects across space while tracking the evolution of various situations over time. Utilizing an autoregressive diffusion transformer, it captures representations related to physics, dynamics, and the principles of cause-and-effect from an extensive collection of visual data, enabling it to simulate environments effectively and develop strategies for managing physical systems. Additionally, Odyssey-3 is capable of creating immersive environments based on user prompts and can instantly predict how these environments shift in response to actions taken by individuals or agents, featuring both first-person and third-person navigation along with flexible camera controls. The extensive world knowledge it acquires can be tailored to various systems by retraining an action decoder or policy using paired data of observations and corresponding actions. Even with limited amounts of task-specific data, Odyssey-3 has demonstrated its utility in controlling robotic arms, powering humanoid robots, operating vehicles, and producing synchronized multi-sensory observations. Furthermore, it is capable of facilitating interactive experiences that enhance user engagement and understanding.

Tencent Yuanbao is an AI-driven assistant that has swiftly gained traction in China, utilizing sophisticated large language models, including its own Hunyuan model, while also integrating with DeepSeek. This application stands out in various domains, such as processing the Chinese language, logical reasoning, and executing tasks efficiently. In recent months, Yuanbao's user base has expanded dramatically, allowing it to outpace rivals like DeepSeek and achieve the top position on the Apple App Store download charts in China. A significant factor fueling its ascent is its seamless integration within the Tencent ecosystem, especially through WeChat, which boosts its accessibility and enhances its array of features. This impressive growth underscores Tencent's increasing ambition to carve out a significant presence in the competitive landscape of AI assistants, as it continues to innovate and expand its offerings. As Yuanbao evolves, it is likely to further challenge existing players in the market.

Marble is an innovative AI model currently undergoing internal testing at World Labs, serving as a variation and enhancement of their Large World Model technology. This web-based service transforms a single two-dimensional image into an immersive and navigable spatial environment. Marble provides two modes of generation: a smaller, quicker model ideal for rough previews that allows for rapid iterations, and a larger, high-fidelity model that, while taking about ten minutes to produce, results in a far more realistic and detailed output. The core value of Marble lies in its ability to instantly create photogrammetry-like environments from just one image, eliminating the need for extensive capture equipment, and enabling users to turn a singular photo into an interactive space suitable for memory documentation, mood board creation, architectural visualization previews, or various creative explorations. As such, Marble opens up new avenues for users looking to engage with their visual content in a more dynamic and interactive way.

NVIDIA Cosmos serves as a cutting-edge platform tailored for developers, featuring advanced generative World Foundation Models (WFMs), sophisticated video tokenizers, safety protocols, and a streamlined data processing and curation system aimed at enhancing the development of physical AI. This platform empowers developers who are focused on areas such as autonomous vehicles, robotics, and video analytics AI agents to create highly realistic, physics-informed synthetic video data, leveraging an extensive dataset that encompasses 20 million hours of both actual and simulated footage, facilitating the rapid simulation of future scenarios, the training of world models, and the customization of specific behaviors. The platform comprises three primary types of WFMs: Cosmos Predict, which can produce up to 30 seconds of continuous video from various input modalities; Cosmos Transfer, which modifies simulations to work across different environments and lighting conditions for improved domain augmentation; and Cosmos Reason, a vision-language model that implements structured reasoning to analyze spatial-temporal information for effective planning and decision-making. With these capabilities, NVIDIA Cosmos significantly accelerates the innovation cycle in physical AI applications, fostering breakthroughs across various industries.

MiniMax M2.5 is a next-generation foundation model built to power complex, economically valuable tasks with speed and cost efficiency. Trained using large-scale reinforcement learning across hundreds of thousands of real-world task environments, it excels in coding, tool use, search, and professional office workflows. In programming benchmarks such as SWE-Bench Verified and Multi-SWE-Bench, M2.5 reaches state-of-the-art levels while demonstrating improved multilingual coding performance. The model exhibits architect-level reasoning, planning system structure and feature decomposition before writing code. With throughput speeds of up to 100 tokens per second, it completes complex evaluations significantly faster than earlier versions. Reinforcement learning optimizations enable more precise search rounds and fewer reasoning steps, improving overall efficiency. M2.5 is available in two variants—standard and Lightning—offering identical capabilities with different speed configurations. Pricing is designed to be dramatically lower than competing frontier models, reducing cost barriers for large-scale agent deployment. Integrated into MiniMax Agent, the model supports advanced office skills including Word formatting, Excel financial modeling, and PowerPoint editing. By combining high performance, efficiency, and affordability, MiniMax M2.5 aims to make agent-powered productivity accessible at scale.

SpAItial is an innovative AI platform dedicated to the creation and implementation of Spatial Foundation Models (SFMs), a groundbreaking category of generative AI systems that excel in generating and interpreting 3D environments while maintaining physical realism and spatial intelligence. Unlike conventional models that independently generate images or text, SpAItial's advanced technology works directly with 3D structures from the beginning, effectively capturing aspects such as geometry, materials, lighting, and physics to create immersive and interactive worlds. Its premier model, Echo-2, possesses the remarkable ability to convert a single image into a fully navigable, photorealistic 3D scene using cutting-edge techniques like Gaussian splatting, which allows users to explore and render environments in real time. This platform is designed with a robust, physically grounded comprehension of space-time, enabling the AI to analyze how objects are situated, interact, and develop within a given environment, eschewing the disjointed outputs typical of traditional generative AI. This innovative methodology not only mitigates the inconsistencies often found in standard generative AI systems but also facilitates a more precise and realistic simulation of environments, paving the way for exciting new applications in virtual reality and beyond.

An advanced End-to-End MLLM is designed to accept various forms of references and effectively ground responses. The Ferret Model utilizes a combination of Hybrid Region Representation and a Spatial-aware Visual Sampler, which allows for detailed and flexible referring and grounding capabilities within the MLLM framework. The GRIT Dataset, comprising approximately 1.1 million entries, serves as a large-scale and hierarchical dataset specifically crafted for robust instruction tuning in the ground-and-refer category. Additionally, the Ferret-Bench is a comprehensive multimodal evaluation benchmark that simultaneously assesses referring, grounding, semantics, knowledge, and reasoning, ensuring a well-rounded evaluation of the model's capabilities. This intricate setup aims to enhance the interaction between language and visual data, paving the way for more intuitive AI systems.

Project Genie is a real-time world generation prototype developed by Google AI. It enables users to create immersive, interactive environments from text descriptions or images. Instead of loading static scenes, Genie generates the world dynamically as you explore. Users can build characters and navigate environments using different movement styles such as walking, flying, or driving. The system supports diverse world types, from photorealistic natural settings to imaginative alien landscapes. Genie responds to user actions with physics, memory, and environmental changes. Each world expands infinitely, creating a sense of continuous exploration. The platform demonstrates how AI can power interactive simulations without prebuilt maps. Project Genie is part of Google’s early research into generative environments. It offers a glimpse into how AI could transform games, simulations, and creative tools.

North Mini Code marks the debut of Cohere’s agentic coding model tailored for developers and serves as the first entry in its next generation of robust models. This compact and efficient open-source solution is specifically crafted for the independent developer community, ensuring remarkable software development capabilities without the need for high-end hardware. Featuring a mixture-of-experts architecture, it comprises a total of 30 billion parameters, with 3 billion of those being active, thereby providing developers with powerful agentic coding functionalities in a streamlined package. The model is finely tuned for various tasks, including code generation, agentic software engineering, and terminal operations, boasting an impressive 256K context length and a maximum generation capacity of 64K. It is designed with real-world developer practices in mind, enabling tasks such as understanding and managing sub-agents, mapping out system architectures, conducting code reviews, and assisting coding agents in navigating intricate software challenges. The integration of these capabilities empowers developers to enhance their productivity and efficiency significantly in software development projects.

HunyuanVideo-Avatar allows for the transformation of any avatar images into high-dynamic, emotion-responsive videos by utilizing straightforward audio inputs. This innovative model is based on a multimodal diffusion transformer (MM-DiT) architecture, enabling the creation of lively, emotion-controllable dialogue videos featuring multiple characters. It can process various styles of avatars, including photorealistic, cartoonish, 3D-rendered, and anthropomorphic designs, accommodating different sizes from close-up portraits to full-body representations. Additionally, it includes a character image injection module that maintains character consistency while facilitating dynamic movements. An Audio Emotion Module (AEM) extracts emotional nuances from a source image, allowing for precise emotional control within the produced video content. Moreover, the Face-Aware Audio Adapter (FAA) isolates audio effects to distinct facial regions through latent-level masking, which supports independent audio-driven animations in scenarios involving multiple characters, enhancing the overall experience of storytelling through animated avatars. This comprehensive approach ensures that creators can craft richly animated narratives that resonate emotionally with audiences.

Aikido Altar represents a cutting-edge open-weight security framework designed to empower organizations with advanced defensive security intelligence tailored for their infrastructures. This model is particularly suited for environments requiring sovereign security where sensitive assets like source code, architectural documents, vulnerability assessments, and other confidential information must remain within the organization's confines and not be shared with external inference services. Built on the GLM-5.3 architecture, Altar employs techniques such as quantization and expert pruning to compress the model size from an extensive 1.51 TB in full precision down to a more manageable 328 GB, all while maintaining the majority of the original model's reasoning and security features. The model retains 168 out of the original 256 routed experts in each backbone expert layer and adopts a W4A16 representation, enhancing its practicality for security tasks that require handling extensive and evolving context windows. The calibration of expert selection was conducted using internal pentesting data and multilingual sources, ensuring that no client information was utilized, which upholds the integrity of cybersecurity, programming, and linguistic capabilities. This innovative approach not only streamlines deployment but also fortifies the organization's security posture against emerging threats.

The GLM-5V-Turbo is an advanced multimodal coding foundation model specifically tailored for tasks that require visual inputs, capable of handling various formats such as images, videos, texts, and files to generate text-based outputs. This model is particularly refined for agent workflows, which allows it to effectively understand environments, plan appropriate actions, and carry out tasks, while also ensuring compatibility with agent frameworks like Claude Code and OpenClaw. Its ability to manage long-context interactions is noteworthy, boasting a context capacity of 200K tokens and an output limit of up to 128K tokens, making it ideal for intricate, long-term projects. Furthermore, it provides a variety of thinking modes suited for diverse scenarios, exhibits robust visual comprehension for both images and videos, and streams output in real-time to enhance user engagement. Additionally, it features sophisticated function-calling abilities that facilitate the integration of external tools, and its context caching capability significantly boosts performance during prolonged conversations. In practical applications, the model can adeptly transform design mockups into fully functional frontend projects, showcasing its versatility and depth in real-world coding scenarios. This versatility ensures that users can tackle a wide range of complex tasks with confidence and efficiency.

Imagen 3 represents the latest advancement in Google's innovative text-to-image AI technology. It builds upon the strengths of earlier versions and brings notable improvements in image quality, resolution, and alignment with user instructions. Utilizing advanced diffusion models alongside enhanced natural language comprehension, it generates highly realistic, high-resolution visuals characterized by detailed textures, vibrant colors, and accurate interactions between objects. In addition, Imagen 3 showcases improved capabilities in interpreting complex prompts, which encompass abstract ideas and scenes with multiple objects, all while minimizing unwanted artifacts and enhancing overall coherence. This powerful tool is set to transform various creative sectors, including advertising, design, gaming, and entertainment, offering artists, developers, and creators a seamless means to visualize their ideas and narratives. The impact of Imagen 3 on the creative process could redefine how visual content is produced and conceptualized across industries.

Game Worlds is an upcoming AI-driven gaming experience developed by Runway, a $3 billion startup that has already made significant impacts on Hollywood with its generative AI technology. The platform currently offers a basic chat interface enabling users to generate text and images, with plans to expand into fully AI-generated video games later this year. Runway’s vision for Game Worlds is to revolutionize the gaming industry by making game development significantly faster and more efficient, similar to AI’s role in accelerating film production. CEO Cristóbal Valenzuela highlights that the gaming sector is adopting AI rapidly, moving faster than Hollywood did two years ago. The platform also intends to collaborate with game companies to train AI models on rich datasets, enhancing its generative capabilities. Game Worlds will provide both gamers and developers with new ways to create, explore, and interact with dynamically generated game content. This initiative is part of Runway’s broader goal to integrate generative AI into creative industries at scale. Game Worlds stands at the forefront of blending AI technology with interactive entertainment.

tresD DICOM offers a complimentary, browser-based platform for viewing dental CBCT and CT scans in various formats such as DICOM, DICOMDIR, ZIP, or RAR, eliminating the need for installation or user accounts. The software operates locally on the user's machine, ensuring that files remain private and are not sent to a remote server for processing. This viewer is equipped with features like 3D volume rendering, the ability to generate multiplanar slices, panoramic images, tooth cross-sections, airway and TMJ visualizations, as well as tools for comparing two CBCT scans. Additionally, users can align intraoral scans saved in STL, PLY, or OBJ formats and can export skull and skin mesh models as STL files. The platform also allows for the creation of simulated cephalometric images and the generation of reports in PDF or PowerPoint formats. Tailored for use by dental professionals and students for educational purposes, the measurements obtained are only indicative, and it is important to note that it is not classified as a medical device. Furthermore, this tool enhances collaboration in dental education by allowing users to easily share visual data with peers and mentors.

Cohere is a robust enterprise AI platform that empowers developers and organizations to create advanced applications leveraging language technologies. With a focus on large language models (LLMs), Cohere offers innovative solutions for tasks such as text generation, summarization, and semantic search capabilities. The platform features the Command family designed for superior performance in language tasks, alongside Aya Expanse, which supports multilingual functionalities across 23 different languages. Emphasizing security and adaptability, Cohere facilitates deployment options that span major cloud providers, private cloud infrastructures, or on-premises configurations to cater to a wide array of enterprise requirements. The company partners with influential industry players like Oracle and Salesforce, striving to weave generative AI into business applications, thus enhancing automation processes and customer interactions. Furthermore, Cohere For AI, its dedicated research lab, is committed to pushing the boundaries of machine learning via open-source initiatives and fostering a collaborative global research ecosystem. This commitment to innovation not only strengthens their technology but also contributes to the broader AI landscape.