Reka AI Labs releases Rho-1 research preview · Launch Video Breakdown: Hook, Pacing & Motion Design
Today, we are releasing a research preview of Rho-1, our 19B omni model that understands and generates text, images, video and robot actions, all inside a single neural network. https://t.co/heADIM5Nt
Scene-by-scene timeline & spoken transcript
The Hook
Anticipatory Countdown & Vision Setting
“Imagine a machine that could think, talk, act, and play.”
- On screen
- 3 Imagine that could think, talk, and play. Wiring models together isn't general intelligence.
- Camera
- Static, centered text with subtle film grain overlay. The initial countdown is a close-up on the number '3'.
- Motion
- Old film countdown leader, typewriter text reveal, subtle film grain overlay, elegant serif typography.
Problem Agitation
The Problem with Current AI Architectures
“(No spoken dialogue — The track opens with a low-frequency hum and a series of electronic beeps, creating a sense of anticipation. A robotic, synthesized voice delivers the core message. This transitions into a high-energy riser at [00:20] followed by a sub-bass drop at [00:22].)”
- On screen
- Wiring models together isn't general intelligence. Fig. 0 · How AI works in 2026 Language model Vision model Image model Video model Robot policy Your request Router Glue code Answer hand-offs 8 elapsed 15.8 s illustrative context re-sent at every hop Today, five models. Each hand-off loses something Language model Vision model Image model Video model Robot model
- Camera
- Static, top-down view of a minimalist flowchart diagram. Text animates in to highlight key components and issues.
- Motion
- Line drawing animation, text reveal, subtle highlight animations, clean diagrammatic presentation.
Product Reveal
Introducing Rho-1: The Unified Network
“(No spoken dialogue — The main musical theme features pizzicato synths, a driving sub-bass, and a consistent 808-style beat.)”
- On screen
- rho-1, one network. rho-1, one network. One set of weights reads and writes every kind of token. understanding generation in out text image video action understanding and generation share attention in every block one state one state, read by every block
- Camera
- Static, centered view of a simplified block diagram. Elements animate to show the unified architecture.
- Motion
- Box and line animation, fill animations, connection line draws, text reveals, highlighting key concepts.
Feature Teaser
Rho-1 in Action: Thinking, Talking, Acting, Playing
“one model that thinks, talks, acts, and plays.”
- On screen
- one model that thinks, rho-1 What happens next in this video? Reason it out. Thought for 2 seconds > The video shows the steam train approaching along the track in the valley. It's still far from the camera, so the next logical be Generating video Continued from your clip + Ask rho-1 anything rho-1 + You're driving down a rho-1 You're driving down a rainy city street at night. What do you watch for? Three short bullet points. • Slippery roads and standing water — reduced tire grip, and watch for oil slicks, puddles, and drain reflections that can hide potholes or debris. • Reduced visibility — rain on your windshield, dimmer streetlights, and oncom rho-1 Third-person video game, a samurai walks across a wooden bridge over a misty river, cherry... 672 x 384 · 24 fps · single pass rho-1 Hands chopping fresh carrots and peppers with a chef's knife on a wooden cutting board in a sunny... 672 x 384 · 24 fps · single pass Live rho-1 streaming session 672 x 384 · 24 fps continuous rho-1, one model that plays,
- Camera
- Dynamic screen recordings of the AI interface, showcasing various capabilities. Quick cuts between different demo scenarios.
- Motion
- Screen capture, UI animation, text typing animation, video playback, grid layout transitions, subtle glow effects.
Call to Action
Reka AI Labs: Rho-1 Research Preview
“(No spoken dialogue — The track builds with additional synth layers and a sustained pad, culminating in a final sub-bass impact at [00:58].)”
- On screen
- ρ-1 RESEARCH PREVIEW reka Labs
- Camera
- Static, centered logo and text reveal over a blurred grid of images.
- Motion
- Fade-in, scale animation, grid background, clean logo reveal.
Motion design primitives
Synthesizable in Remotion · key components of the unified-ai-launch-template template.
TypewriterTextReveal
Text appears character by character, mimicking a classic typewriter effect, often accompanied by a subtle cursor animation.
+ Code construction recipe
Use `interpolate` to animate the `clipPath` or `width` of a text layer, combined with a blinking cursor component. Apply a `spring` for the cursor's blink animation.
DiagrammaticFlowBuild
Boxes and connecting lines animate into view sequentially, illustrating a system or process flow, with text labels appearing alongside.
+ Code construction recipe
Utilize `Sequence` and `AbsoluteFill` components. Animate `opacity` and `scale` for boxes, and `strokeDashoffset` for lines using `interpolate` with `easeOut` for a smooth build-up.
DynamicUIMockup
Simulated user interface elements, such as text input fields, chat bubbles, and video players, animate to demonstrate product functionality.
+ Code construction recipe
Layer `AbsoluteFill` components for UI elements. Use `spring` for button presses or text field focus, `interpolate` for typing animations (similar to TypewriterTextReveal), and `Video` components for embedded media playback.
Style DNA & aesthetic system
Visual aesthetic
Typography
Motion
Special effects
Pacing & rhythm
Conceptual story rhythm
- 1
The Enigmatic Question
Purpose. To immediately capture attention and establish a grand, aspirational vision for AI.
Execution. A vintage film countdown leads into elegant serif text appearing character by character on a dark, grainy background, posing a profound question about intelligent machines.
- 2
The Current State & The Gap
Purpose. To articulate the limitations of existing AI architectures and set up the problem that the new product solves.
Execution. Minimalist, animated flowcharts on a light background illustrate the complexity and inefficiencies of 'wiring models together', highlighting 'hand-offs' and 'lost context'.
- 3
The Elegant Solution
Purpose. To introduce Rho-1 as a paradigm shift, a unified and elegant solution to the identified problem.
Execution. A clean, simplified block diagram animates to show 'rho-1, one network', with elements highlighting its single-network, multi-modal capabilities, accompanied by a rising musical theme.
- 4
The Live Demonstration
Purpose. To provide tangible proof of Rho-1's capabilities across various modalities (text, image, video, action).
Execution. Rapid-cut screen recordings showcase the AI interacting with different prompts and generating diverse outputs, from reasoning about video to creating game environments and controlling robots, all within a clean UI.
Remix in Animatiq Studio
Master remix prompt · agent-ready
Sample remix
Objective
Launch a new quantum computing simulation platform, 'QubitForge', that unifies disparate simulation tools into a single environment.How the recipe adapts
Remix blueprint · unified-ai-launch-template
Scene durations (5 scenes, 60s)
Key Remotion components
Related AI Agents Product Launches
Explore all AI Agents launches →
To ensure that artificial general intelligence benefits all of humanity
Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use. Its capabilities exceed those of any model we’ve ever made generally available. https://t.co/2AvmEjHIX8
To understand the true nature of the universe.