Runway AI Cheatsheet
A comprehensive guide to Runway's AI features, capabilities, and pricing as of May 2025
Runway ML is an applied AI research company focused on advancing creativity through artificial intelligence, offering advanced tools for art, entertainment, and human expression. The platform specializes in generative AI models that help transform ideas into visual reality with minimal technical knowledge required.
Core Technology Offerings
General World Models
AI systems that understand and generate comprehensive visual environments
Multi-Motion Brush
Sophisticated motion control in generated videos
Gen-4
Their latest and most powerful generative AI system (released April 2025)
NEWLatest Updates (as of May 2025)
Runway's latest features are available to all subscription tiers, but may use different amounts of credits depending on your plan. Always check the official Runway blog for the very latest announcements.
Runway offers several cutting-edge AI models with varying capabilities, each specialized for different creative tasks.
Gen-4 NEW
Released April 2025, Gen-4 is Runway's most advanced video generation system.
- Character Consistency: Maintains consistent characters across different scenes and shots
- Coherent Environments: Creates stable, persistent world environments throughout videos
- Perspective Control: Can regenerate elements from different angles within scenes
- Visual References: Utilizes reference images combined with text instructions
- Enhanced Control: Features like advanced camera paths and improved ControlNet integration.
Gen-3 Alpha
High-quality video generation model with improved fidelity and control.
- Video Quality: Improved fidelity, consistency, and motion compared to previous models
- Video Length: Supports up to 40-second videos through extensions
- Camera Control: Options for static camera and handheld camera shake effects
- Restyle Feature: Ability to apply visual styles from reference images to videos.
Frames
Advanced image generation model for high-fidelity still images.
- High-Fidelity Images: Advanced base model for detailed image creation
- Stylistic Control: Superior control over visual details and aesthetic
- Reference Processing: Convert reference images to text prompts and custom styles
Model Feature Comparison
| Feature | Gen-4 | Gen-3 Alpha | Frames |
|---|---|---|---|
| Character Consistency | ✓✓✓ | ✓✓ | N/A |
| Environmental Coherence | ✓✓✓ | ✓✓ | N/A |
| Multi-Motion Control | ✓✓✓ | ✓✓ | N/A |
| Advanced Camera Control | ✓✓✓ | ✓ | N/A |
| Text Prompt Complexity | ✓✓✓ | ✓✓ | ✓✓✓ |
| Reference Image Use | ✓✓✓ | ✓✓ | ✓✓✓ |
Runway provides comprehensive tools for generating, editing, and enhancing video content through AI.
Video Creation Methods
Text-to-Video
Generate videos directly from text descriptions with support for various styles and cinematic controls.
Image-to-Video
Transform static images into dynamic videos with first frame control and detailed text prompting.
Video-to-Video
Apply different styles to existing videos while preserving motion and structure, including the Restyle feature.
Video Editing Features
- 4K Resolution: Upscale video to 4K for professional output
- Extension: Extend clips beyond original length (up to 40 seconds with Gen-3, potentially longer or higher quality with Gen-4)
- Expand Video: Transform videos into new aspect ratios
- Editing Tools: Trim, reverse, adjust video speed
- Background Removal: Isolate subjects from backgrounds
- Color Grading: Text-based color adjustment
Special Effects and Controls
For best results with video generation, start with a clear main subject in your prompt, then add visual details, camera specifications, lighting information, and style references. Utilize features like Multi-Motion Brush and Camera Paths to refine your output.
Runway's image tools allow for high-fidelity image creation with precise stylistic control.
Frames Image Model
Advanced base model for high-fidelity image creation with superior control.
- Advanced base model for high-fidelity image creation
- Superior stylistic control and visual detail
- Reference image conversion to text prompts and custom styles
Image Capabilities
Frames can:
- Generate images up to 1024×1024 resolution (and potentially higher with upscaling tools)
- Accept text prompts with detailed descriptions
- Use reference images as style guides or structural input
- Maintain coherent visual aesthetics
- Create diverse artistic styles and realistic images
When creating images to use as video generation references (e.g., for Image-to-Video or style references), focus on clear composition with defined foreground subjects and uncluttered backgrounds for optimal results.
Runway provides various audio generation and transformation tools to complement its visual features.
Audio Features
Text-to-Speech
- Convert text to natural-sounding voice
- Support for up to 10,000 characters
- Multiple voice options and styles
Speech-to-Speech
- Retain content, pace, and tone while changing voice
- Natural voice transformation
Lip Sync
- Match facial movements to audio tracks
- Support for multiple faces simultaneously
- Videos up to 45 seconds in length
Custom Voices
Runway allows you to create personalized voices from short audio samples (available in Pro plan and higher):
- Create personalized voices from short audio samples
- Share with workspace collaborators
- Requires explicit user consent for creation
Runway offers several pricing tiers to accommodate different needs and usage levels.
| Plan | Price | Credits | Key Features |
|---|---|---|---|
| Free | $0 | 125 one-time |
|
| Standard | $15/mo $12/mo (yearly) | 625 monthly |
|
| Pro | $35/mo $28/mo (yearly) | 2,250 monthly |
|
| Unlimited | $95/mo $76/mo (yearly) | Higher allocation (e.g., 10,000+) or truly unlimited for certain tasks. |
|
| Enterprise | Custom (starts ~$1500/year) | Custom amounts |
|
Note: Pricing, credit allocations, and specific feature availability per plan are subject to change. Always refer to the official RunwayML website for the most up-to-date and detailed information.
Credit System Explained
Credits are the currency used within Runway ML to generate content:
- Different generation types consume varying amounts of credits (e.g., Gen-4 video costs more than Gen-3)
- Higher quality outputs, longer videos, and advanced features use more credits
- Credits refresh monthly based on your subscription plan or can be purchased as top-ups.
For teams or frequent users, the Pro or Unlimited plans offer the best balance of features and credit allocation. Individuals starting out may find the Standard plan sufficient, but should monitor credit usage with newer models like Gen-4.
Effective prompting is key to getting the best results from Runway's AI models.
Do's
- Be descriptive: Focus on visual details rather than abstract concepts. Describe what you want to see.
- Think cinematically: Describe what the "camera" is capturing, including shot types, angles, and movement.
- Use technical terms: Include camera specifications (e.g., "35mm lens," "wide angle"), lens types, lighting terms ("golden hour," "volumetric lighting").
- Keep it direct: Use simple, clear language that describes the scene. Be specific.
- Specify art style: (e.g., "impressionistic painting," "sci-fi concept art," "photorealistic").
- Iterate: Start simple and add complexity. Refine prompts based on results.
Don'ts
- Avoid conversational language: (e.g., "Can you please make me a video...")
- Avoid command-based prompts for generation: (e.g., "Add a dog to this clip..." - use editing tools for this). For generation, describe the scene *with* the dog.
- Don't use overly negative phrasing: (e.g., "No clouds in the sky..." - instead describe "clear blue sky"). Some negative prompts are supported but positive description is usually better.
- Don't provide multi-shot instructions in one prompt: One prompt generally creates one continuous shot or image.
- Avoid overly complex or contradictory descriptions in a single prompt.
Expert Prompt Structure
A good starting structure for prompts, especially for video:
[SUBJECT] + [ACTION/SETTING] + [VISUAL DETAILS/MOOD] + [CAMERA SHOT/ANGLE/MOVEMENT] + [LIGHTING] + [STYLE]
Example: "A majestic dragon soaring through a stormy sky, lightning illuminates its scales, dramatic clouds, aerial shot tracking the dragon, low angle view, rim lighting, fantasy art style, cinematic."
When working with Gen-4, focus on describing the world, characters, and their actions clearly. Gen-4 excels at maintaining consistency, so detailed descriptions of characters and environments can lead to more coherent and believable results across scenes and perspectives.