Loading playground
Input
Additional Settings
Customize your input with more control.
Result
Idle
Input
Additional Settings
Customize your input with more control.
This model is no longer supported.
Customize your input with more control.
Customize your input with more control.
Customize your input with more control.
Ovis Image's 7B architecture delivers specialized text rendering at $0.012 per megapixel. Trading general-purpose image generation for typography-focused accuracy, this model solves the persistent problem of legible, aesthetically integrated text in AI-generated visuals. Purpose-built for designers and developers who need clean text overlays without post-processing.
Use Cases: Marketing Graphics with Typography | UI Mockups with Text Elements | Social Media Posts with Captions
At $0.012 per megapixel, Ovis Image delivers specialized text rendering capabilities at roughly 3x the cost of general-purpose alternatives, justified by eliminating manual text correction workflows.
| Metric | Result | Context |
|---|---|---|
| Architecture Size | 7B parameters | Optimized specifically for text rendering vs general 10B+ models |
| Inference Speed | 3-5 seconds | Standard acceleration mode on fal infrastructure |
| Cost per Megapixel | $0.012 | 83 megapixels per $1.00 on fal |
| Max Resolution | 1024x768 (landscape_4_3) | Multiple aspect ratios available via image_size parameter |
| Batch Generation | 1-4 images | Cost scales linearly per image at $0.012/MP each |
Most text-to-image models struggle with legible text integration; letters blur, words scramble, or typography feels disconnected from the visual composition. Ovis Image's 7B architecture prioritizes text coherence and aesthetic integration over photorealistic detail.
What this means for you:
Clean typography rendering: Generate marketing graphics, social posts, or presentation slides with readable text embedded directly in the image, no Photoshop cleanup required
Flexible acceleration modes: Choose between regular (balanced) or high (faster) acceleration based on whether you're iterating concepts or producing finals
Controlled inference: 1-50 step range with guidance scale 1-20 lets you dial in the exact balance between prompt adherence and creative interpretation
Safety-first deployment: Built-in content filtering (enable_safety_checker) returns NSFW flags per image, critical for client-facing or public applications
| Spec | Details |
|---|---|
| Architecture | Ovis Image 7B |
| Input Formats | Text prompts, negative prompts, seed control |
| Output Formats | JPEG, PNG, WebP |
| Max Resolution | 1024x768 (multiple aspect ratios via image_size) |
| License | Commercial use permitted |
API Documentation | Quickstart Guide | Enterprise Pricing
AuraFlow Text to Image ($0.004/MP) – Ovis Image ($0.012/MP) trades cost efficiency for specialized text rendering accuracy at 3x the price. AuraFlow prioritizes general-purpose image quality and faster generation for workflows where embedded text isn't critical, making it ideal for concept art, illustrations, or photorealistic scenes without typography requirements.
