What is AskGL? The Command Language for AI Power Users
While Ask gives you conversational AI assistants, and Fabrics give you form-based tools, AskGL gives you something different: direct control.
AskGL is Promptha's expressive command language for AI generation. Think of it as a domain-specific language (DSL) designed for power users who want precise control over AI models without clicking through menus or filling out forms.
One line of AskGL can do what takes multiple steps elsewhere:
/image @fal.flux-schnell $landscape !4k --repeat 3 cyberpunk cityscape at night
This single command generates three 4K landscape images of a cyberpunk cityscape using Fal's Flux Schnell model. No forms. No menus. Just keyboard.
Table of Contents
- Why AskGL Exists
- The 28 Commands
- Model Routing with @provider.model
- References: Context from Anywhere
- Parameters, Presets, and Modifiers
- Multi-Model Execution
- AskGL vs Ask vs Fabric
- Real-World Examples
Why AskGL Exists
Every AI interface makes trade-offs. Chatbots are flexible but require explanation. Forms are guided but constrained. Both are mouse-heavy and menu-driven.
AskGL optimizes for a different use case: experienced users who know what they want and want to get there fast.
If you're generating images all day, you don't want to:
- Click a dropdown to select a model
- Navigate tabs to set dimensions
- Toggle switches for quality settings
- Wait for form validation
You want to type a command and hit enter.
AskGL gives you:
- Speed: Type faster than you can click
- Precision: Control every parameter explicitly
- Reproducibility: Save commands as templates
- Composability: Chain models together
- Flexibility: Mix and match any syntax elements
It's the difference between using a GUI and using a command line. Both have their place. AskGL is for when you want the command line.
The 28 Commands
AskGL includes 28 commands organized by function:
Generation Commands
| Command | What it does |
|---|---|
/image | Generate images from text descriptions |
/video | Create video clips from prompts |
/audio | Generate speech, music, sound effects |
/3d | Create 3D models |
Image Transforms
| Command | What it does |
|---|---|
/edit | Modify existing images |
/vary | Create variations of an image |
/upscale | Increase image resolution (up to 8K) |
/remove-bg | Remove image backgrounds |
/extend | Extend/outpaint images |
/remix | Apply new styles to images |
Text Generation
| Command | What it does |
|---|---|
/write | Generate written content |
/summarize | Condense content |
/expand | Elaborate on content |
/rewrite | Change style or tone |
/outline | Create structured outlines |
/translate | Convert between languages |
Data & Structure
| Command | What it does |
|---|---|
/table | Generate formatted tables |
/json | Output structured JSON |
/csv | Generate CSV data |
/extract | Pull structured data from text |
/diagram | Create diagrams (Mermaid format) |
Code
| Command | What it does |
|---|---|
/code | Generate code snippets |
/explain | Explain code or concepts |
/debug | Debug code issues |
/convert | Convert between languages |
Business
| Command | What it does |
|---|---|
/slides | Generate presentation content |
/email | Write professional emails |
/report | Generate reports |
Analysis
| Command | What it does |
|---|---|
/analyze | Analyze content |
/caption | Generate image captions |
/compare | Compare items |
Each command has sensible defaults, but you can override everything with the syntax elements below.
Model Routing with @provider.model
AskGL's most powerful feature is direct model routing. Instead of selecting from a dropdown, you specify exactly which model to use:
@provider.model-id
Available Providers
| Provider | Specialty | Example Models |
|---|---|---|
@fal | Images, video, audio, 3D | flux-schnell, flux-pro, veo-3.1, maya-tts |
@replicate | Community marketplace | ideogram-v3, sora-2 |
@openai | GPT family, DALL-E | gpt-4o, gpt-4o-mini |
@anthropic | Claude family | claude-sonnet-4, claude-opus-4 |
@google | Gemini, Imagen | gemini-2.5-pro, imagen-4 |
@deepseek | Efficient models | deepseek-chat |
@elevenlabs | Text-to-speech | Voice models |
Examples
/image @fal.flux-schnell mountain landscape
/write @anthropic.claude-sonnet-4 technical documentation
/video @fal.veo-3.1 cinematic drone shot over forest
/audio @fal.maya-tts professional voiceover script
Smart Defaults
If you don't specify a model, AskGL picks intelligently based on:
- The command type (
/image→ image model,/write→ LLM) - Quality modifiers (
!fast→ fast models,!4k→ premium models) - Your preferences and history
Explore all available models in LLM Models on Promptha.
References: Context from Anywhere
The @ syntax isn't just for models. You can reference content from across your workspace:
| Syntax | What it references |
|---|---|
@prompt:name | Saved prompt template |
@kb:name | Knowledge base content |
@out:name | Named previous output |
@file:name | Uploaded file |
@last | Most recent output |
@prev:N | Nth previous output |
@style:name | Built-in style preset |
Example with References
/image @fal.flux-schnell @kb:brand-colors @out:logo-v1 marketing banner
This command:
- Uses Flux Schnell for generation
- Pulls your brand colors from your knowledge base
- References your existing logo as visual context
- Generates a marketing banner incorporating all of it
No copy-pasting. No re-uploading. Just reference and go.
Parameters, Presets, and Modifiers
Parameters ($key:value)
Fine-tune generation with explicit parameters:
| Parameter | Purpose | Example |
|---|---|---|
$width:N | Width in pixels | $width:1920 |
$height:N | Height in pixels | $height:1080 |
$aspect:W:H | Aspect ratio | $aspect:16:9 |
$duration:N | Video/audio length | $duration:10 |
$seed:N | Reproducible results | $seed:12345 |
$steps:N | Inference steps | $steps:50 |
$guidance:N | Guidance scale | $guidance:7.5 |
$fps:N | Video frame rate | $fps:24 |
Presets ($name)
Common configurations in one word:
| Preset | Dimensions | Use Case |
|---|---|---|
$cinematic | 21:9 aspect | Widescreen shots |
$portrait | 9:16 (1080x1920) | Mobile content |
$landscape | 16:9 (1920x1080) | Desktop content |
$square | 1:1 (1024x1024) | Social posts |
$youtube-thumbnail | 16:9 (1280x720) | YouTube thumbnails |
$instagram-story | 9:16 (1080x1920) | Stories |
$twitter-header | 3:1 (1500x500) | Twitter headers |
Modifiers (!flag)
Quick quality and behavior adjustments:
| Modifier | Effect |
|---|---|
!fast | Prioritize speed |
!hd | 1080p output |
!4k | 4K resolution |
!8k | 8K ultra resolution |
!loop | Create looping video |
!seamless | Tileable output |
Flags (--flag value)
Execution control:
| Flag | Purpose | Example |
|---|---|---|
--repeat N | Generate N variations | --repeat 4 |
--seed N | Set seed for reproducibility | --seed 42 |
--ref @output | Reference for chaining | --ref @out:v1 |
Multi-Model Execution
This is where AskGL gets powerful. You can chain models together or run them in parallel.
Chained Execution (|)
Pass output from one model to the next:
/image @fal.flux-schnell | @fal.crystal-upscaler concept art character
This generates an image with Flux Schnell, then passes it to Crystal Upscaler for enhancement. The output of each model becomes the input of the next.
Use cases:
- Generate → Upscale
- Generate → Remove Background
- Draft → Refine → Polish
Parallel Execution (,)
Run multiple models simultaneously and compare:
/image @fal.flux-schnell, @fal.ideogram-v3, @google.imagen-4 product shot
This runs all three models in parallel. You get three outputs to compare. Pick the best one, or use all three.
Use cases:
- A/B testing models
- Generating variations
- Finding the best model for a task
Limits
- Maximum 3 chain steps (A | B | C)
- Maximum 4 parallel models per step
- Maximum 6 total operations per expression
Combined Example
/image @fal.flux-schnell, @fal.ideogram-v3 | @fal.crystal-upscaler
$4k product photography
This runs Flux and Ideogram in parallel, then upscales both results.
AskGL vs Ask vs Fabric
All three are AI interfaces on Promptha. Here's when to use each:
| Aspect | AskGL | Ask | Fabric |
|---|---|---|---|
| Interface | Command line | Chat | Form |
| Model selection | @provider.model syntax | Dropdown or automatic | Pre-configured |
| Multi-model | Chains and parallel | No | No |
| Learning curve | Higher | Low | Lowest |
| Speed (expert) | Fastest | Medium | Medium |
| Customization | Infinite | Moderate | Defined by creator |
| Best for | Power users, automation | Exploration, conversation | Defined tasks |
Use AskGL when:
- You know exactly what you want
- You're doing repetitive generation tasks
- You want to chain models together
- You need precise parameter control
- You're building automation
Use Ask when:
- You want to explore through conversation
- You need domain-specific context
- Requirements aren't fully defined
- You prefer chat over commands
Use Fabric when:
- You have a specific, defined task
- You want guided inputs with validation
- You need consistent, reproducible outputs
- You're not comfortable with syntax
Real-World Examples
Marketing Asset Generation
/image @fal.flux-schnell @kb:brand-guidelines $landscape !4k
--repeat 3 hero banner for product launch
Generate three 4K hero banners using your brand guidelines.
Content Workflow
/write @anthropic.claude-sonnet-4 @kb:tone-guide
blog post about AI productivity tips
/image @fal.flux-pro $landscape !hd
blog header image for AI productivity article
Write the post, then generate a matching header image.
Video Thumbnail Pipeline
/image @fal.flux-schnell | @fal.crystal-upscaler
$youtube-thumbnail !4k exciting thumbnail for coding tutorial
Generate a thumbnail and upscale it to 4K in one command.
Multi-Model Comparison
/image @fal.flux-schnell, @fal.ideogram-v3, @google.imagen-4
$square minimalist logo concept for tech startup
Test three different models to see which produces the best logo.
Professional Voiceover
/audio @fal.maya-tts $duration:30
professional female voice for product explainer video
Generate a 30-second professional voiceover.
Getting Started with AskGL
1. Start Simple
Begin with basic commands:
/image sunset over mountains
/write email to thank a client
AskGL will use smart defaults. Once comfortable, add more syntax.
2. Learn the Presets
Presets eliminate parameter lookup:
/image $portrait profile photo
/image $youtube-thumbnail video thumbnail
/video $cinematic establishing shot
3. Try Model Routing
Specify models when you want control:
/image @fal.flux-schnell fast draft
/image @fal.flux-pro final version
4. Build Chains
Once familiar, chain operations:
/image @fal.flux-schnell | @fal.crystal-upscaler draft to final
5. Save Templates
Save frequently-used commands as templates for one-click access.
What's Next?
Now that you understand AskGL, explore further:
- LLM Models on Promptha - All available text models
- Image Generation Models - Available image models
- What is Ask? - Conversational AI alternative
- What is a Fabric? - Form-based AI tools
- What is Knowledge Base? - Context for your commands
AskGL is designed for users who think in commands, not clicks. Once you learn the syntax, it becomes the fastest way to control AI generation on Promptha. Start with simple commands, build up to chains, and you'll find yourself reaching for AskGL whenever speed and precision matter.