Thank you for providing the detailed file. Here is a comprehensive list of the newest Fal AI endpoints for image and video generation, image upscaling, and guidance on image-to-prompt, along with full example code for each, as documented in your latest README_fal_ai.md file.

🏆 Latest Fal AI Endpoints & Full Usage Code

🖼️ Image Generation Endpoints

1. Google Imagen 4 (2025)

Endpoint: fal-ai/imagen4/preview

import { fal } from '@fal-ai/client';

const result = await fal.subscribe('fal-ai/imagen4/preview', {
  input: {
    prompt: 'A hyper-detailed photorealistic cityscape at sunset',
    image_size: '1024x1024'
  }
});

console.log('Generated image:', result.data.images[0].url);

2. FLUX.1 Pro Ultra

Endpoint: fal-ai/flux-pro/v1.1-ultra

import { fal } from '@fal-ai/client';

const result = await fal.subscribe('fal-ai/flux-pro/v1.1-ultra', {
  input: {
    prompt: 'Professional headshot, studio lighting',
    image_size: '1536x1536'
  }
});

console.log('Generated image:', result.data.images[0].url);

3. FLUX.1 [dev]

Endpoint: fal-ai/flux/dev

import { fal } from '@fal-ai/client';

const result = await fal.subscribe('fal-ai/flux/dev', {
  input: {
    prompt: 'A photorealistic portrait of a person',
    image_size: '1024x1024',
    num_inference_steps: 28
  }
});

4. FLUX.1 [schnell] (Fastest)

Endpoint: fal-ai/flux/schnell

import { fal } from '@fal-ai/client';

const result = await fal.subscribe('fal-ai/flux/schnell', {
  input: {
    prompt: 'A quick sketch of a cat',
    num_inference_steps: 4
  }
});

5. FLUX with LoRA

Endpoint: fal-ai/flux-lora

const result = await fal.subscribe('fal-ai/flux-lora', {
  input: {
    prompt: 'A portrait in anime style',
    loras: [{
      path: 'https://url-to-lora-weights.safetensors',
      scale: 1.0
    }]
  }
});

6. SANA 4K Generator

Endpoint: fal-ai/sana (Usage pattern similar to above endpoints.)

🎬 Video Generation Endpoints

1. Google Veo 3 (2025)

Endpoint: fal-ai/veo3

const result = await fal.subscribe('fal-ai/veo3', {
  input: {
    prompt: 'A person walking down a busy street with city sounds',
    duration: 5, // in seconds
    include_audio: true
  }
});

Note: Supports video with native audio generation.

2. Stable Video Diffusion

Endpoint: fal-ai/stable-video-diffusion

const result = await fal.subscribe('fal-ai/stable-video-diffusion', {
  input: {
    prompt: 'A sunrise timelapse in a mountain landscape',
    duration: 4
  }
});

🖌️ Image Editing & Upscaling Endpoints

1. FLUX Kontext with LoRA

Endpoint: fal-ai/flux/kontext-lora (For advanced image editing.)

2. FLUX Differential Diffusion

Endpoint: fal-ai/flux-differential-diffusion (For special effects and differential edits.)

3. Background Removal

Endpoint: fal-ai/birefnet

const result = await fal.subscribe('fal-ai/birefnet', {
  input: {
    image_url: 'https://example.com/photo.png'
  }
});

4. Image Upscaling (ESRGAN)

Endpoint: fal-ai/esrgan

const result = await fal.subscribe('fal-ai/esrgan', {
  input: {
    image_url: 'https://example.com/photo.png'
  }
});

5. Face Swap

Endpoint: fal-ai/face-swap (For swapping faces between images.)

🧠 Image to Prompt (Image Captioning)

As of the information in your file, Fal AI does not offer a dedicated "image-to-prompt" endpoint yet. However, you can watch for updates on the official Fal AI models page as new captioning endpoints may be introduced in 2025.

✅ Best Practices for Using Fal AI Endpoints

Use fal.subscribe(endpoint, { input: {...} }) for quick, interactive tasks.
For long-running or queued jobs (especially large images or training), use the queue pattern:

// Submit a task
const { request_id } = await fal.queue.submit('endpoint-name', { input: { ... } });

// Check job status
const status = await fal.queue.status('endpoint-name', { requestId: request_id });

// Get the result
const result = await fal.queue.result('endpoint-name', { requestId: request_id });

📦 Installation & Setup

npm install @fal-ai/client

import { fal } from '@fal-ai/client';
fal.config({ credentials: 'your_api_key_here' });

🌐 Official Resources & Documentation

🔔 Recent Major Updates (2024–2025)

May 2025: Google Veo 3 video+audio, Imagen 4 improvements
December 2024: ElevenLabs TTS, Resemble AI, FLUX Kontext, open LoRA weights
November 2024: FLUX Pro Ultra (2K resolution)

⚡ Summary Table

Endpoint	Type	Released	Use Case
`fal-ai/imagen4/preview`	Image Generation	2025	Hyper-detailed images (Google)
`fal-ai/flux-pro/v1.1-ultra`	Image Generation	2024	Pro 2K images
`fal-ai/flux/dev`	Image Generation	2024	Dev, best quality
`fal-ai/flux/schnell`	Image Generation	2024	Fastest
`fal-ai/veo3`	Video Generation	2025	Video (w/ audio)
`fal-ai/stable-video-diffusion`	Video Generation	2024	High-quality video
`fal-ai/esrgan`	Image Upscaling	Ongoing	2x/4x upscaling

For the full endpoint JSON, see fal_ai_endpoints.json in your repository for schema and more examples. For detailed, always-updated reference, check Fal AI Models.

If you need sample code for a specific endpoint or use-case (e.g., inpainting, prompt-to-video, LoRA, etc.), let me know! All code here is from your latest Fal AI README and matches the current (2025) API.

svngoku/fal-2025-api.md

Select an option

No results found