Imagen 4 · Fast image previewinference · 2.9s

Imagen 4 · Fast

Image·imagen4·by Google

Use this fast version of Imagen 4 when speed and cost are more important than quality

Try it now →
API reference
Runtime (p50)
13s
Estimated price
$0.02
Call the API
prediction.sh
sh
curl -X POST \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -H "Content-Type: application/json" \
  --data '{
    "model": "imagen-4-fast",
    "version": "0.0.1",
    "input": {
        "prompt": "Create a cinematic photorealistic medium shot capturing the intense atmosphere of a high speed coastal highway at golden hour. Slightly off center stands a young man with wind blown dark hair sun kissed skin and a piercing focused gaze locked directly into the camera. His expression is calm yet determined with a subtle tension in his jaw. He wears a fitted storm gray and burnt orange racing jacket unzipped to clearly reveal a matte black t shirt that prominently displays the text Imagen 4 Fast in bold stylized motion lettering across the chest with on Eachlabs written just beneath it in clean modern typography both lines fully visible and unobstructed. A pair of aviator sunglasses rests naturally on his head glinting softly in the low angle sunlight. The background features a windswept coastal cliff road with a blurred motion trail of a speeding car curving into the distance. The ocean sparkles far below while warm golden highlights from the setting sun cast long dramatic shadows. The composition balances natural beauty and tension with high contrast lighting drawing crisp attention to the man face and torso while the background remains soft and slightly blurred. Every element from the movement in his hair to the clarity of the printed text reinforces a grounded high impact cinematic realism.",
        "aspect_ratio": "1:1",
        "output_format": "jpg",
        "safety_filter_level": "block_only_high"
    },
    "webhook_url": ""
}' \
  https://api.eachlabs.ai/v1/prediction/
Documentation8 sections
  • Overview

    imagen-4-fast — Text-to-Image AI Model

    imagen-4-fast delivers rapid photorealistic image generation from text prompts, prioritizing speed and low cost for developers and creators needing quick prototypes without sacrificing core quality. Developed by Google as part of the imagen4 family, this text-to-image AI model generates images in about 4 seconds at just 2 credits per output, making it ideal when Google text-to-image performance matters more than ultra-high fidelity. Unlike full-quality variants, imagen-4-fast optimizes for low-latency use cases like iterative prompt testing or high-volume text-to-image AI model workflows on platforms like Eachlabs.

  • Capabilities

    Photorealism: Create lifelike images of people, animals, landscapes, and more—down to the finest detail.

    Ultra-sharp details: Rich textures, vibrant colors, and stunning close-ups with natural depth and gradients.

    Smarter text rendering: Better spelling, longer passages, and more sophisticated typography—ideal for comics, collectibles, and design work.

    More styles, more control: From hyper-realistic to abstract, it handles a wide range of visual styles with improved accuracy.

    Fast mode: Explore ideas at lightning speed—up to 10× faster than Google’s earlier models.

    High-resolution output: Generate crisp, creative visuals at up to 2K resolution.

  • Use cases

    Use Cases for imagen-4-fast

    Developers building text-to-image APIs can leverage imagen-4-fast for rapid prototyping, generating dozens of variants in minutes to refine app features before scaling to premium models. Input a prompt like "photorealistic close-up of a leather wallet on a wooden table with soft sunlight," and receive a high-fidelity 1024x1024 output in seconds for instant feedback.

    Marketers creating Google text-to-image content for e-commerce benefit from its speed in producing lifestyle product images, such as swapping backgrounds on inventory photos without studio delays, ensuring quick campaign visuals at low cost.

    Content creators testing prompts for social media use imagen-4-fast's low-latency edge to iterate on aspect ratios like 9:16 for stories, generating photorealistic portraits or scenes far faster than full-quality alternatives.

    Designers in high-volume workflows appreciate its texture accuracy for fabric and material mocks, enabling fast approvals on previews before final renders.

  • Tips & tricks

    How to Use imagen-4-fast on Eachlabs

    Access imagen-4-fast seamlessly on Eachlabs via the Playground for instant text-to-image testing, the API for scalable integrations, or SDK for custom apps—simply provide a descriptive prompt, select aspect ratio and resolution up to 1024x1024, and get photorealistic outputs in ~4 seconds with full commercial rights.

    ---
  • Technical spec

    What Sets imagen-4-fast Apart

    imagen-4-fast stands out in the text-to-image landscape with its balance of photorealistic detail and blazing speed, generating 1024x1024 images in roughly 4 seconds—half the time of the standard Imagen 4 at a fraction of the cost. This enables developers seeking a Google text-to-image API to handle prototyping or batch tasks efficiently without premium pricing.

    • Ultra-low latency at ~4 seconds per image: Perfect for real-time testing of prompts in text-to-image AI model apps, where standard models take 8+ seconds, allowing 2x faster iteration cycles.
    • 1024x1024 resolution with flexible aspect ratios (1:1, 16:9, 9:16, 4:3, 3:4): Supports common formats for social media or web previews directly, reducing post-processing needs compared to lower-res alternatives.
    • Photorealistic rendering of textures and fabrics at 2 credits: Excels in product visuals and lifestyle shots where fine details matter, outperforming slower models in cost-sensitive high-volume scenarios.

    These specs make imagen-4-fast a go-to for fast text-to-image generation, distinct from higher-res but slower competitors.

  • Things to be aware of

    No customization tools: It doesn’t support style transfer, subject tuning, or few-shot personalization.

    No image editing: Features like inpainting, outpainting, masking, or image upscaling are not available.

    No negative prompting: You can’t exclude elements (e.g., “no text,” “no watermark”) via prompts.

  • Key considerations

    Prompt quality matters: Clear, detailed prompts lead to better results.

    Not fact-grounded: Expect some inconsistencies in details and realism.

    Style & text limits: Handles diverse styles and text better—but not perfectly.

  • Limitations

    Lack of factual grounding: Imagen isn’t built for real-world accuracy. It can introduce artifacts in complex scenes, especially

    with small faces, text, or thin structures.

    Centering issues: Struggles with perfect alignment, such as placing a circle exactly in the center.

    Unclear prompts: Nonsensical input (like random characters or emojis) can lead to unpredictable results.

    Output Format: JPG,PNG

Related models

4 models
Flux 2 Klein 4B Base · Text to Image AI model preview
Flux 2 Klein 4B Base · Text to ImageBlack Forest Labs
Wan v2.6 · Text to Image AI model preview
Wan v2.6 · Text to ImageAlibaba
Recraft v4 · Text to Image AI model preview
Recraft v4 · Text to Imagerecraft
Flux 2 Klein 9B Base · Text to Image AI model preview
Flux 2 Klein 9B Base · Text to ImageBlack Forest Labs
* FAQ

About Imagen 4 · Fast

01 / 03

What is Google Imagen 4 Fast and how does it generate images?

Imagen 4 Fast is Google's speed-optimized version of the Imagen 4 text-to-image model, designed to deliver fast image generation at lower cost while maintaining strong visual quality. It generates detailed, photorealistic images from natural language prompts using Google's advanced diffusion architecture, optimized for workflows requiring high throughput.