Mintplex-Labs/anything-llm
1
0
Fork 0
mirror of https://github.com/Mintplex-Labs/anything-llm.git synced 2026-07-21 01:25:28 -04:00
Code Issues 328 Packages Projects Releases 10 Wiki Activity

[GH-ISSUE #943] [FEAT]: Image generation #573

New Issue
Open
opened 2026-02-22 18:20:14 -05:00 by yindo · 17 comments
yindo commented 2026-02-22 18:20:14 -05:00
Owner
Copy Link
Copy Source

Originally created by @Mirgiacomo on GitHub (Mar 21, 2024).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/943

Originally assigned to: @timothycarambat on GitHub.

What would you like to see?

In addition to text generation, is it possible to allow anythingLLM to generate images?
Via the image generation API of OpenAI, or Gemini, etc.?
PS: i mean text to image and not image to text

Originally created by @Mirgiacomo on GitHub (Mar 21, 2024). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/943 Originally assigned to: @timothycarambat on GitHub. ### What would you like to see? In addition to text generation, is it possible to allow anythingLLM to generate images? Via the image generation API of OpenAI, or Gemini, etc.? PS: i mean text to image and not image to text
yindo added the enhancementIntegration Requestfeature request labels 2026-02-22 18:20:14 -05:00
yindo commented 2026-02-22 18:20:14 -05:00
Author
Owner
Copy Link
Copy Source

@ytjhai commented on GitHub (Sep 5, 2024):

I'm also interested in this, specifically around the way images can be "retrieved" from existing documentation rather than generated from scratch using something like DALL-E. Embedding images is currently possible with some LLMs but some things can only be communicated through the original image. So, retrieving that image in its original form would help a lot. The "Show Citations" modal would be the part where this can be visible.

@ytjhai commented on GitHub (Sep 5, 2024): I'm also interested in this, specifically around the way images can be "retrieved" from existing documentation rather than generated from scratch using something like DALL-E. Embedding images is currently possible with some LLMs but some things can only be communicated through the original image. So, retrieving that image in its original form would help a lot. The "Show Citations" modal would be the part where this can be visible.
yindo commented 2026-02-22 18:20:14 -05:00
Author
Owner
Copy Link
Copy Source

@appdevhm88mt commented on GitHub (Oct 23, 2024):

Hi, regarding the function of generating images from text, is there any estimate of when or which version it will be supported by?
Thx

@appdevhm88mt commented on GitHub (Oct 23, 2024): Hi, regarding the function of generating images from text, is there any estimate of when or which version it will be supported by? Thx
yindo commented 2026-02-22 18:20:15 -05:00
Author
Owner
Copy Link
Copy Source

@severfire commented on GitHub (Jan 28, 2025):

Its something very much needed. Embedding images and generating images or links to them.

@severfire commented on GitHub (Jan 28, 2025): Its something very much needed. Embedding images and generating images or links to them.
yindo commented 2026-02-22 18:20:15 -05:00
Author
Owner
Copy Link
Copy Source

@keramblock commented on GitHub (Feb 14, 2025):

also Flux.1 would be cool

@keramblock commented on GitHub (Feb 14, 2025): also Flux.1 would be cool
yindo commented 2026-02-22 18:20:15 -05:00
Author
Owner
Copy Link
Copy Source

@morbificagent commented on GitHub (Mar 21, 2025):

Oh yes, would be very great!
@timothycarambat in #3386 you say that there isnt a model that can be run local and gives good quality on low end devices... i understand that but using an online model would be great anyway :-)

@morbificagent commented on GitHub (Mar 21, 2025): Oh yes, would be very great! @timothycarambat in #3386 you say that there isnt a model that can be run local and gives good quality on low end devices... i understand that but using an online model would be great anyway :-)
yindo commented 2026-02-22 18:20:15 -05:00
Author
Owner
Copy Link
Copy Source

@psimm commented on GitHub (Apr 4, 2025):

My team and I would appreciate image generation via an external API, e.g. a FLUX or Stable Diffusion model.

@psimm commented on GitHub (Apr 4, 2025): My team and I would appreciate image generation via an external API, e.g. a FLUX or Stable Diffusion model.
yindo commented 2026-02-22 18:20:15 -05:00
Author
Owner
Copy Link
Copy Source

@Arp1tMoga commented on GitHub (Apr 21, 2025):

+1

@Arp1tMoga commented on GitHub (Apr 21, 2025): +1
yindo commented 2026-02-22 18:20:15 -05:00
Author
Owner
Copy Link
Copy Source

@timothycarambat commented on GitHub (Apr 21, 2025):

please dont comment +1

We cannot track comments of +1 - leave a reaction so that it gets bumped!
Filter we use: https://github.com/Mintplex-Labs/anything-llm/issues?q=is%3Aissue+is%3Aopen+sort%3Areactions-desc+

@timothycarambat commented on GitHub (Apr 21, 2025): _please dont comment +1_ We cannot track comments of `+1` - leave a reaction so that it gets bumped! Filter we use: https://github.com/Mintplex-Labs/anything-llm/issues?q=is%3Aissue+is%3Aopen+sort%3Areactions-desc+
yindo commented 2026-02-22 18:20:15 -05:00
Author
Owner
Copy Link
Copy Source

@acuentascanarias commented on GitHub (Apr 22, 2025):

We have the same need as @Mirgiacomo . Is it possible to allow anythingLLM to generate images?
I mean text to image (image to text already available)

@acuentascanarias commented on GitHub (Apr 22, 2025): We have the same need as @Mirgiacomo . Is it possible to allow anythingLLM to generate images? I mean text to image (image to text already available)
yindo commented 2026-02-22 18:20:15 -05:00
Author
Owner
Copy Link
Copy Source

@IronBeardKnight commented on GitHub (May 19, 2025):

Having ability to use even localai's ability to do image generating, tts etc would be very cool

@IronBeardKnight commented on GitHub (May 19, 2025): Having ability to use even localai's ability to do image generating, tts etc would be very cool
yindo commented 2026-02-22 18:20:15 -05:00
Author
Owner
Copy Link
Copy Source

@dimdieo commented on GitHub (Jun 24, 2025):

It would be great if they made support for Comfyui

@dimdieo commented on GitHub (Jun 24, 2025): It would be great if they made support for Comfyui
yindo commented 2026-02-22 18:20:15 -05:00
Author
Owner
Copy Link
Copy Source

@cannin commented on GitHub (Sep 30, 2025):

@timothycarambat Below is a simple MCP server to get an image from an external source that could be used for image generation. Using @agent will recognize the commands. The problem is that the tiny image (star.png) is returned and shown, but not the large one (star_big.png); larger images always seem to take longer and get severely cutoff. Not sure if it is a AnythingLLM or Python FastMCP problem.

Image

MCP Config

From: anythingllm_mcp_servers.json

{
  "mcpServers": {
    "getImages": {
      "type": "streamable", 
      "url": "http://127.0.0.1:8000/mcp"
    }
  }
}

MCP Server

#!/usr/bin/env -S uv run --script
# /// script
# requires-python = ">=3.10"
# dependencies = [
#     "fastmcp",
#     "requests",
#     "Pillow",
# ]
# ///
import requests
import io
import base64
from PIL import Image as PILImage

from fastmcp import FastMCP
from fastmcp.utilities.types import Image
from mcp.types import ImageContent

mcp = FastMCP("Get Images")


@mcp.tool
def get_external_image() -> Image:
    """Retrieve the external montage image and return it as an image"""
    url = "https://raw.githubusercontent.com/cannin/cannin.github.io/refs/heads/master/images/star_big.png"
    response = requests.get(url)

    if not response.ok:
        return "Failed to download the external image."

    buffer = io.BytesIO(response.content)
    return Image(data=buffer.getvalue(), format="png")


@mcp.tool
def get_network_image() -> Image:
    """
    Download the network image and return it as an image.
    """
    url = "https://raw.githubusercontent.com/cannin/cannin.github.io/refs/heads/master/images/star.png"
    response = requests.get(url)

    if not response.ok:
        raise Exception("Failed to download the network image.")

    buffer = io.BytesIO(response.content)
    return Image(data=buffer.getvalue(), format="png")


if __name__ == "__main__":
    mcp.run(transport="http", host="127.0.0.1", port=8001, path="/mcp")

Related to: https://github.com/Mintplex-Labs/anything-llm/issues/1526

@cannin commented on GitHub (Sep 30, 2025): @timothycarambat Below is a simple MCP server to get an image from an external source that could be used for image generation. Using @agent will recognize the commands. The problem is that the tiny image (star.png) is returned and shown, but not the large one (star_big.png); larger images always seem to take longer and get severely cutoff. Not sure if it is a AnythingLLM or Python FastMCP problem. <img width="780" height="1356" alt="Image" src="https://github.com/user-attachments/assets/c6e0cc3a-47fb-4485-97f4-3490197417c4" /> ## MCP Config From: anythingllm_mcp_servers.json ``` { "mcpServers": { "getImages": { "type": "streamable", "url": "http://127.0.0.1:8000/mcp" } } } ``` ## MCP Server ``` #!/usr/bin/env -S uv run --script # /// script # requires-python = ">=3.10" # dependencies = [ # "fastmcp", # "requests", # "Pillow", # ] # /// import requests import io import base64 from PIL import Image as PILImage from fastmcp import FastMCP from fastmcp.utilities.types import Image from mcp.types import ImageContent mcp = FastMCP("Get Images") @mcp.tool def get_external_image() -> Image: """Retrieve the external montage image and return it as an image""" url = "https://raw.githubusercontent.com/cannin/cannin.github.io/refs/heads/master/images/star_big.png" response = requests.get(url) if not response.ok: return "Failed to download the external image." buffer = io.BytesIO(response.content) return Image(data=buffer.getvalue(), format="png") @mcp.tool def get_network_image() -> Image: """ Download the network image and return it as an image. """ url = "https://raw.githubusercontent.com/cannin/cannin.github.io/refs/heads/master/images/star.png" response = requests.get(url) if not response.ok: raise Exception("Failed to download the network image.") buffer = io.BytesIO(response.content) return Image(data=buffer.getvalue(), format="png") if __name__ == "__main__": mcp.run(transport="http", host="127.0.0.1", port=8001, path="/mcp") ``` Related to: https://github.com/Mintplex-Labs/anything-llm/issues/1526
yindo commented 2026-02-22 18:20:15 -05:00
Author
Owner
Copy Link
Copy Source

@manuelkamp commented on GitHub (Oct 19, 2025):

any progress or update on this feature?
I'd really like to use image generation with my localai in anythingllm too (since I use anythingllm as my primary ai tool). now I have to use nextcloud assistant for image generations :)

@manuelkamp commented on GitHub (Oct 19, 2025): any progress or update on this feature? I'd really like to use image generation with my localai in anythingllm too (since I use anythingllm as my primary ai tool). now I have to use nextcloud assistant for image generations :)
yindo commented 2026-02-22 18:20:15 -05:00
Author
Owner
Copy Link
Copy Source

@ChupaUps commented on GitHub (Feb 2, 2026):

Following up on previous requests about image generation support. Any progress on multi-modal outputs (images/files) from LLMs? Currently using external tools for this while Anything LLM remains my primary AI tool. Looking for updates or workarounds.

@ChupaUps commented on GitHub (Feb 2, 2026): Following up on previous requests about image generation support. Any progress on multi-modal outputs (images/files) from LLMs? Currently using external tools for this while Anything LLM remains my primary AI tool. Looking for updates or workarounds.
yindo changed title from [FEAT]: Image generation to [GH-ISSUE #943] [FEAT]: Image generation 2026-06-05 14:36:04 -04:00
yindo commented 2026-06-05 14:36:04 -04:00
Author
Owner
Copy Link
Copy Source

@jjnxpct commented on GitHub (Mar 4, 2026):

I recently started using AnythingLLM and I love the platform, but I was surprised to find that image generation is not yet supported. This is something I was really hoping to offer my clients.

I tried working around this by building a custom agent skill that calls the BFL Flux API, and the skill itself works correctly — but I can't get it to function reliably because I exclusively want to use EU-based models (Mistral). Mistral models consistently fail to pass arguments to the tool call, so the skill is invoked but receives empty parameters every time.

It would be great to see native image generation support added — ideally with the ability to connect models like Flux (BFL) for generation and editing. This would be a natural fit for a platform with such strong agent capabilities.

Hoping to see this on the roadmap soon.

<!-- gh-comment-id:3998823008 --> @jjnxpct commented on GitHub (Mar 4, 2026): I recently started using AnythingLLM and I love the platform, but I was surprised to find that image generation is not yet supported. This is something I was really hoping to offer my clients. I tried working around this by building a custom agent skill that calls the BFL Flux API, and the skill itself works correctly — but I can't get it to function reliably because I exclusively want to use EU-based models (Mistral). Mistral models consistently fail to pass arguments to the tool call, so the skill is invoked but receives empty parameters every time. It would be great to see native image generation support added — ideally with the ability to connect models like Flux (BFL) for generation and editing. This would be a natural fit for a platform with such strong agent capabilities. Hoping to see this on the roadmap soon.
yindo commented 2026-06-05 14:36:05 -04:00
Author
Owner
Copy Link
Copy Source

@DDesmond95 commented on GitHub (Apr 23, 2026):

Still not supported?

<!-- gh-comment-id:4304037867 --> @DDesmond95 commented on GitHub (Apr 23, 2026): Still not supported?
yindo commented 2026-06-05 14:36:05 -04:00
Author
Owner
Copy Link
Copy Source

@LeighBicknell commented on GitHub (May 3, 2026):

Also wanting support for this.

<!-- gh-comment-id:4365617180 --> @LeighBicknell commented on GitHub (May 3, 2026): Also wanting support for this.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
Desktop
Docker
Integration Request
Integration Request
OS: Linux
OS: Mobile
OS: Windows
UI/UX
blocked
bug
bug
core-team-only
documentation
duplicate
embed-widget
enhancement
feature request
github_actions
good first issue
investigating
needs info / can't replicate
possible bug
pull-request
Mirrored from GitHub Pull Request
 question
stage: specifications
wontfix
No Label Integration Request enhancement feature request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Mintplex-Labs/anything-llm#573
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?