Imference – Roadmap & Updates

Follow the evolution of the Imference image generation API. This page lists what we’re building next and the ideas we’re exploring for more advanced users.

Want to start using the API now? Go to the main API page →

What’s coming next

We keep Imference intentionally small and focused: a fast API, simple credits, and a curated set of models. Below are the next planned improvements, plus one unplanned idea we’re exploring.

Base64 image responses

PLANNED

Add an option to receive generated images directly as base64 in the JSON response, instead of (or in addition to) the current image URL.

  • Easier integration in “no file storage” environments.
  • Useful for previews, browser-only flows, or custom storage.

New model support

PLANNED

Extend the model field to support additional, state-of-the-art models.

  • flux1.1-pro – Flux Pro 1.1 for high-quality, flexible generations.
  • gpt-image-1 – OpenAI’s GPT Image-1 model.
  • All new models will use the same credits system.

Advanced generation controls

IDEA (UNPLANNED)

Explore more advanced request fields for power users. This is not scheduled yet, but under active consideration.

  • Batch generation (multiple images per request).
  • Custom negative text (negative words / prompts).
  • CFG / guidance scale control.
  • Choice of scheduler for diffusion steps.

How it will look in the API

These changes are designed to be backwards compatible. Existing integrations will continue to work as-is, while new options will be available for users who want more control.

1. Optional base64 output (planned)

We plan to add a flag in the /image/generate request body to request base64-encoded images: 

POST /image/generate
Content-Type: application/json
Authorization: Bearer <API_KEY>

{
  "model": "illustrious-hassuka-xl",
  "prompt": "A cat astronaut floating in space",
  "return_base64": true
}

The response will keep the current structure (URLs) and optionally include a base64 field, so you can migrate at your own pace.

2. New model codes (planned)

Once enabled, you’ll be able to use the new models by setting the model field: 

{
  "model": "flux1.1-pro",
  "prompt": "Cinematic portrait of a person standing in a neon-lit city at night"
}
{
  "model": "gpt-image-1",
  "prompt": "Isometric illustration of a modern workspace with multiple screens"
}

No pricing changes are planned at this time: 1 credit still equals 1 generated image, any model.

3. Advanced generation controls (idea, not scheduled)

If we move forward with advanced controls, they would likely be additional, fully optional fields on /image/generate. A possible shape could look like: 

POST /image/generate
Content-Type: application/json
Authorization: Bearer <API_KEY>

{
  "model": "illustrious-hassuka-xl",
  "prompt": "A detailed fantasy castle on a cliff",
  "negative_prompt": "blurry, low resolution, watermark",
  "batch_size": 4,
  "guidance_scale": 7.5,
  "scheduler": "euler_a"
}

The exact field names and supported values may change, or this feature might not ship at all. If/when it becomes planned, we’ll update the status on this page.

Stay up to date

We announce new features and breaking changes ahead of time. You can periodically check this page, or keep an eye on our main site for important notices.