Skip to content
Images

SIMB: A Cross-Modal Speech and Vision AI Model

1 min readupdated

Developed by the KU-AGI team, the SIMB model combines the visual expertise of Janus-Pro-7B with a speech expert based on Qwen2.5-3B. It handles cross-modal synthesis tasks including speech-to-image generation, image-to-speech conversion, and spoken visual question answering (spoken VQA). The base visual expert remains frozen and identical to the original, while specialized bridge modules connect the different modalities. This new release is ideal for researchers and developers working on multimodal systems.

Key features

  • Support for generating images from audio recordings
  • Reverse conversion of images into speech tokens
  • Ability to ask questions about images and receive spoken answers
  • Fully standalone repository that requires no third-party weight downloads

What you can do

  • Generate an image from an audio prompt using the built-in code
  • Add voice output to visual content within multimodal pipelines
  • Build an experimental voice assistant with image comprehension

Example

from simb import SIMB   # pip install -e . in [https://github.com/KU-AGI/SIMB](https://github.com/KU-AGI/SIMB)
model = SIMB.from_pretrained("KU-AGI/SIMB")
image, _ = model.generate_image("prompt.wav", seed=0)
Read moreDetails
Total parameters: 11.8B
Weight format: bf16 (for main components)
Base models: Janus-Pro-7B, Qwen2.5-3B, DualCodec, w2v-BERT 2.0
License: combined (requires compliance with DeepSeek, Qwen, and MIT terms)
Discuss Telegram 0

Will it run?

Runs in FP8 / INT8

Full precision does not fit, a quantised version does: quality barely changes down to 8 bits, more noticeably at 4 bits.

  • BF16weights 22 GB · ready files26 GB
  • FP8 / INT8recommendedweights 11 GB14 GB
  • 4-bitweights 6.2 GB8.3 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 5080 · BF16 — from RTX 5090

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # KU-AGI/SIMB
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download KU-AGI/SIMB --local-dir ~/models/SIMB
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to SIMB
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • SageAttention 3v2.2.0from October 28, 2025

    even faster on RTX 50 / Blackwell

    FP4 attention for Blackwell. Faster than SageAttention 2, but the authors suggest v2 when accuracy matters.

    How to enable: Build from the sageattention3_blackwell folder of the repo; in ComfyUI via KJNodes.

    Earlier generations (2)
    • SageAttention 2++ — Quantised (INT8/FP8) attention with no visible quality loss. The main speed-up for image and video generation.
    • SageAttention 1 — First version: INT8 for Q·K. Still works, but 2++ is faster and more accurate.
  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (3)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
    • xFormers — The older way to speed up attention, works almost everywhere. Sage or Flash are faster where available.
  • cache-dit (DBCache)v1.5.6from November 20, 2026new release

    1.5–3× at nearly the same quality

    Reuses block outputs between diffusion steps. Supports most DiT models.

    How to enable: pip install cache-dit; one line cache_dit.enable_cache(pipe) for diffusers.

    Earlier generations (2)
    • MagCache — Skips steps by magnitude of change. More accurate than TeaCache, no calibration.
    • TeaCache — The first popular diffusion step cache. Works, newer methods are more accurate.
  • Nunchaku (SVDQuant 4-bit)v1.2.1from January 25, 2026

    ≈3× faster and 3.5× less memory

    4-bit versions of popular image models with almost no quality loss (FP4 on RTX 50).

    How to enable: Ready weights for supported models; in ComfyUI use ComfyUI-nunchaku.

  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of Janus-Pro-7B (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 4All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 8All on Hugging Face

About the model

KU-AGI/SIMB
Author
KU-AGI
Category
Images
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments

Other

OmniSRE

OmniSRE
Language models

OnyX-9B

OnyX-9B