Image Generation — AI Draws Pictures

Imagine telling a chef “make me a pizza with pineapple and extra cheese,” and having it appear on your plate before you finish the sentence. That’s the magic Hermes Agent brings to image creation—type a description, and a picture materializes in under a second.

Meet Your Artists

Hermes Agent comes with 11 built-in AI painting engines, each with its own personality. Some are speed demons, some are perfectionists, and they all charge different rates. Here are the headliners:

Model	Speed	Superpower	Price
flux-2/klein/9b (default)	<1 sec	Crisp text, lightning fast	~$0.006/megapixel
flux-2-pro	~6 sec	Studio-grade photorealism	~$0.03/megapixel
ideogram/v3	~5 sec	Typography & art lettering	$0.03–0.09/image
krea/v2/large	25–60 sec	Cinematic film grain	$0.06–0.065/image

Plain talk: The default model is perfect for everyday quick sketches. Need a commercial poster? Switch to recraft/v4/pro. Want anime-style illustrations? Grab krea/v2/medium.

Step 1: Get Your API Key (Free Trial)

1. Sign up at fal.ai
2. Generate an API key in your dashboard (it looks like a random string of letters and numbers)

Pro tip: Already paying for Nous Portal? Skip this entirely—use the built-in “Tool Gateway” and save your wallet.

Step 2: Choose Your Artist

Open your terminal and run:

hermes tools

You’ll see an interactive menu. Use arrow keys to select 🎨 Image Generation, then pick your backend (FAL.ai or Nous Subscription). A model list appears:

  Model                          Speed    Strengths                    Price
  fal-ai/flux-2/klein/9b         <1s      Fast, crisp text             $0.006/MP   ← current
  fal-ai/flux-2-pro              ~6s      Studio photorealism          $0.03/MP
  fal-ai/z-image/turbo           ~2s      Bilingual EN/CN, 6B          $0.005/MP

Press up/down to browse, Enter to select. Your choice saves automatically to config.yaml.

Step 3: Peek at Your Settings (Optional)

Want to tweak things manually? Open config.yaml:

image_gen:
  model: fal-ai/flux-2/klein/9b   # current model
  use_gateway: false              # true = use Nous subscription
  max_parallel_requests: 4        # max simultaneous generations

The max_parallel_requests setting controls how many images generate at once. Default is 4, and the system keeps it reasonable automatically.

Final Thoughts & Practical Advice

The core idea: Hermes Agent wraps 11 AI painting models into simple commands—pick one, then just describe what you want in plain words.

My honest recommendation: Start with the default flux-2/klein/9b. It’s fast, cheap, and perfect for testing ideas. When you need something portfolio-worthy (posters, product shots), upgrade to recraft/v4/pro or gpt-image-2—costs more, but the quality speaks for itself. And if a model throws an HTTP 4xx error? Don’t panic—it’s probably temporarily unavailable. Switch models or set your FAL_KEY and move on.

Now go describe your first masterpiece. The canvas is waiting.