SenseNova 6.8 Flash
More Reliable Workflows. Better Design.
Model Overview
SenseNova 6.8 Flash is an AI agent model designed for complex real-world tasks, office productivity, and multimodal creation.
Operating as a goal-oriented collaborator, SenseNova 6.8 Flash supports complex, multi-step tasks and collaboration across tasks and sessions, with a focus on reliable execution and delivering usable results.
With long-term memory and proactive interaction, the model can connect historical context with new developments, identify potential risks, and help move tasks forward across sessions.
Beyond task execution, SenseNova 6.8 Flash extends its generative capabilities into AI-native design, supporting websites, SVG graphics, interactive Web 3D, dynamic presentations, animated posters, and editable Blender projects.
Its goal is not simply to generate content, but to deliver useful, actionable, and editable results.
Key Capabilities
- Complex Task Execution: Supports complex, multi-step workflows and collaboration across tasks and sessions, with a focus on reliable task execution and delivering usable results.
- Long-Term Memory and Proactive Interaction: Maintains context around user goals and constraints, identifies affected tasks when information changes, and suggests adjustments to keep work moving forward.
- AI-Native Design and Dynamic Creation: Supports web design, SVG, interactive Web 3D, dynamic PPT, animated posters, and iterative Blender 3D creation, extending generation into practical and editable deliverables.
办公智能
Long-Term Memory and Proactive Interaction
SenseNova 6.8 Flash explores a more proactive approach to office intelligence. With long-term memory and proactive interaction, the model can maintain context across tasks and sessions, connect historical information with new developments, identify potentially affected tasks, and suggest adjustments to help work move forward.
Its long-term interaction capabilities are evaluated across factual recall, instruction following, conflict resolution, and memory of user preferences and goals.
Design Capabilities
Design Capabilities
Leveraging its multimodal capabilities, SenseNova 6.8 Flash extends design generation across a wide range of creative formats, including Web, SVG, Three.js, Blender, and PPT, supporting diverse AI-native design needs. It also enables dynamic presentations, animated posters, and agentic 3D design.
Additionally, the model can integrate with workflows such as Figma, helping users bring AI-powered design capabilities into their existing professional creative processes.
Dynamic PPT
SenseNova 6.8 Flash goes beyond adding animations to slides by incorporating narrative pacing into the generation process. It coordinates slide transitions, element animations, and visual layouts to deliver cohesive, dynamic PPTX presentations.
Dynamic Poster
Dynamic Poster goes beyond simply adding motion to static visuals. SenseNova 6.8 Flash coordinates hero visuals, text hierarchy, motion timing, and visual rhythm in a unified generation process. The resulting content can be used for social media campaigns and extended into a consistent series of visual assets.
View prompt
We would like you to create a dynamic poster for our theatre production of 'MONOLITH' using the provided high-contrast stage photograph. Please incorporate the title 'MONOLITH' along with the descriptive text 'A Dystopian Chamber Drama', the billing 'Directed by Clara Vance', and the run details 'The Warehouse Mainstage | Oct 12-28'. In your design, make sure to emphasize the monumental geometry of the massive suspended concrete-textured blocks and the lone silhouette beneath them to capture the dystopian scale and atmosphere of the show.
Blender 3D
SenseNova 6.8 Flash can work with Blender through an agent to complete complex 3D creation. It generates an initial scene, reviews rendered results, responds to feedback, and iteratively refines the output into an editable Blender project.
View prompt
Create a refined, approximately 20-second animation of a premium vinyl record player. The centerpiece is an original high-end desktop turntable. It features a smoked walnut plinth with softly rounded edges, a graphite-colored top plate, understated champagne-gold hardware, and dark isolation feet. A precision-machined aluminum platter supports a deep-black vinyl record with a warm-white center label. A slim silver tonearm features a dark cartridge, visible stylus, counterweight, and precision pivot assembly. Use an elegant three-quarter side view, soft showroom lighting, and a warm, light-colored background. Emphasize rich material textures, restrained highlights, and crisp, well-defined edges. At the beginning, the vinyl record and platter float in separate layers above the solid plinth, revealing the small motor, rubber drive belt, and central bearing beneath. The tonearm assembly hovers slightly above its mounting base. After a brief pause, each component gradually settles into its proper assembled position. The tonearm base connects to the top plate, and the platter smoothly accelerates into rotation. The tonearm first moves over the record with the stylus raised, then gently lowers, allowing the stylus tip to make realistic contact with the outer groove. In the final shot, the record spins steadily as the tonearm slowly tracks a short distance inward along the groove, settling into an elegant playback position. Use the center label and subtle surface details to make the rotation clearly visible. Throughout the animation, the plinth, spindle, bearing, platter, record, tonearm pivot, and stylus must remain structurally consistent and belong to the same coherent turntable assembly.
Web Design
Supports a wide range of web design needs, combining page layouts, typography, color palettes, and visual assets to create cohesive designs across diverse styles, from minimalist branding and creative visuals to tech-focused aesthetics.
SVG Design
Supports a wide range of vector graphics and animation needs, combining shape construction, color schemes, path control, and motion sequencing to create cohesive, scalable, and editable vector designs.
Web 3D
Creates real-time 3D content rendered in the browser. The model supports 3D scene creation and interactive experiences, handling spatial layout, materials and lighting, camera controls, and animation sequencing. Web 3D turns scenes into browser-based interactive content, allowing users to rotate viewpoints, move cameras, trigger interactions, and explore immersive experiences.
View prompt
Create an aurora lakeside research camp with a warm, cartoon-like aesthetic using Three.js. The scene should combine the structural detail of a miniature architectural diorama with the spacious atmosphere of a winter night. In the foreground, place a naturally contoured frozen lake, with its shoreline edged by snowdrifts, a few scattered rocks, and sparse snow-covered cedar trees. Several low-rise cabins are arranged irregularly along the far side of the lake, forming a small observation compound. The main cabin, duty station, and storage shed feature soft gray-green, dark red, or natural wood-colored walls. Their roofs are covered with naturally varying layers of snow, while doors, windows, chimneys, eaves, and entrance steps should all have clearly defined proportions and dimensions. Behind the observation compound, build an elevated circular observation platform supported by solid columns or an enclosed structural base, with a lightweight safety railing around its perimeter. Install a telescope and antenna on the platform, both angled toward the sky. Clearly model the connections between the telescope tube, mounting bracket, pivot joints, and base. The instruments should be appropriately sized relative to the available standing space. Add a spiral staircase along the side of the observation platform. Its steps should wind around a central axis, with continuous handrails, connecting to the platform entrance at the top and reaching the snowy path leading to the duty station at the bottom. Leave sufficient walking clearance between the cabins, staircase, and platform. A short wooden bridge extends from the shoreline in front of the main cabin toward a sampling hut in the middle of the lake. Clearly model the bridge deck, railings, and supporting piles. The sampling hut should have structural supports extending down into the lakebed. Include a small ice hole and a simple sampling rig beside the hut. The frozen lake surface should feature fine cracks, frosted patches, and subtle layers of semi-transparent ice. Place an equipment shed toward the rear side of the camp, with a small snowmobile parked underneath. Neatly arrange equipment cases and coiled cables nearby without obstructing the snowy paths. Footprints and snowmobile tracks should follow realistic walking and driving routes throughout the camp. Leave ample open space in the background for the night sky. Several flowing curtains of turquoise-green and pale blue aurora should slowly unfold across the sky, with continuous variations in brightness and folds. Their soft green reflections should gently illuminate the surface of the frozen lake. Use cool blue tones as the dominant palette for the sky and snow, contrasting with the warm, inviting light glowing from inside the cabins. Low-level lights along the shoreline should illuminate only nearby entrances and snowy paths. Set the default camera to an elevated oblique angle, capturing both the camp and the lake while keeping the distant mountains and aurora visible. Support camera rotation and zoom. Maintain clear visual depth and separation between the snow-covered ground, frozen lake, wooden cabins, and night sky. The overall atmosphere should feel peaceful, cozy, and lived-in.
Design Workflows
Integrates with professional design tools such as Figma, bringing AI-powered design capabilities into designers' existing workflows. It connects content generation, editing and refinement, and iterative design, enabling smoother collaboration between AI-assisted creation and designers' professional expertise.
设计能力
从 Vibe Coding 到 Vibe Design
依托多模态能力,SenseNova 6.8 聚焦设计场景,逐步完善覆盖多种设计场景与工作流的能力体系。覆盖 Web、SVG、Three.js、Blender、PPT 等多种创作形态,并在业内率先支持动态 PPT、动态 Poster 以及 Agentic 3D 动态创作:
基础设计能力
网页与矢量设计
网页设计
Web 设计方面,支持多样化的网页设计需求,结合页面布局、字体排版、色彩搭配与视觉素材,生成风格统一的网页设计,呈现从简约品牌风到创意视觉、科技产品风的多样化表达。
SVG 设计
SVG 设计方面,支持多样化的矢量图形与动画设计需求,结合图形构造、色彩搭配、路径控制与动效编排,生成风格统一、可缩放和可编辑的矢量设计。
进阶设计能力
复杂 3D 创作
3D 复杂建模设计
3D 方面,支持多样化的三维场景与交互设计需求,结合空间布局、材质光照、镜头控制与动画编排,构建可在网页中实时渲染的三维内容,呈现从立体视觉展示到动态交互的多样化表达。
设计工作流
接入 Figma 专业流程
支持接入 Figma 等专业设计工具,将 AI 设计能力融入设计师现有工作流,衔接内容生成、编辑调整与方案迭代,支持 AI 辅助创作与设计师精修协同。
动态创作
从 Vibe Design 到 Vibe Dynamics
动态 PPT
动态 Poster
动态 3D Blender 设计
Suitable Audiences
Suitable Audiences
- Application Developers and Agent Engineers: Building multi-step task workflows involving tool use, structured outputs, reasoning, and end-to-end task delivery
- Enterprise Teams and Office Professionals: Exploring AI-assisted project coordination, business opportunity tracking, knowledge continuity, and proactive workflow support
- Designers and Dynamic Content Creators: Creating websites, vector graphics, presentations, animated posters, and other visual content
- Web Creators and 3D Professionals: Developing interactive browser-based 3D experiences and editable Blender scenes
Experience
Experience
1. SenseNova Studio
- Open SenseNova Studio: https://unify.light-ai.top/
- Sign in using your GitHub account or Google account
- Click “Design Beta” in the navigation menu
- Enter an invitation code to access the feature. Please contact the official SenseNova team to obtain an invitation code
Available Design Capabilities
| Category | Access |
|---|---|
| Dynamic PPT | Available in SenseNova Studio |
| Website | Available in SenseNova Studio |
| Dynamic Poster | Available in SenseNova Studio; supports uploading reference images |
For 3D capabilities, we recommend using SenseNova Token Plan.
2. SenseNova Token Plan
Related Research
Related Research | Exploring Next-Generation Model Architectures
SenseTime has also open-sourced SenseNova Loop-9B-Preview, a 9B-scale Looped Transformer model exploring next-generation language model architectures.
The project investigates how parameter-shared recurrent computation can expand effective computational depth without increasing the parameter count, alongside techniques such as Manifold-Constrained Hyper-Connections (mHC) to improve information flow and stability.
This research is a separate architectural exploration and should not be confused with the SenseNova 6.8 Flash official release.