Skip to main content
Bring images into your ongoing conversations without switching endpoints. In-chat Image Generation lets you request images directly from the chat/completions endpoint by selecting a base chat model and an image generation model using a simple suffix syntax.

​
How it works

  • Base model: Your chat model (e.g., claude-sonnet-4-20250514)
  • Image model: Your image generator (e.g., flux-pro)
  • Suffix syntax: baseModel@imageModel
Example suffix: claude-sonnet-4-20250514@flux-pro When you use a suffixed model with chat/completions, the assistant can respond with content that includes generated image(s) inline with its text, keeping the flow immersive for roleplay and storytelling scenarios.

​
Example request

curl https://api.electronhub.ai/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4-20250514@flux-pro",
    "messages": [
      {"role": "system", "content": "You are a narrative game master. Embed images when helpful."},
      {"role": "user", "content": "Generate a vivid, cinematic portrait of a cyberpunk detective at night in neon rain."}
    ]
  }'
Notes:
  • The base chat model drives conversation quality and reasoning.
  • The image model controls visual style and fidelity.
  • The response may include image URLs or Markdown image formatting inside the assistant message content.

​
SillyTavern example

Here’s how this looks in SillyTavern when using a suffixed model: SillyTavern in‑chat image generation

​
Tips

  • Be descriptive: Specify subject, style, lighting, composition, and mood.
  • Stay consistent: Reuse key style terms across turns for continuity.
  • Guide placement: Ask the assistant where to insert images within the story.