Cerberus E4B needs about 7.2 GB of memory at Q4_K_M (a 5.3 GB file) with an 8K context window, so it fits on a 8 GB GPU. This repository has 4 quants, from 5.3 GB (Q4_K_M) to 15.1 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 5.6 GB.
Cerberus E4B – 7.52B multimodal LLM with vision, audio, and reasoning
Cerberus E4B is a 7.52B parameter model based on gemma-4-E4B-it. It supports text, image, and audio input and generates text output. The model is available in several GGUF quantizations for local deployment.
Good at
- Multimodal reasoning
- Tool usage
- Audio and image processing
Limits
- No vision support in some quantizations
- Large VRAM needed for higher precision
- License is Apache‑2.0
Best for
- Local chatbots with image input
- Audio transcription and analysis
- Reasoning tasks on laptops
Who it’s for. Developers and researchers who need a compact, multimodal LLM for on‑device inference. The model is suitable for applications that require text, image, and audio processing without cloud dependence.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 5.3 GB | 7.2 GB | Tight | Get file → | |
| Q5_K_M | 5.8 GB | 7.7 GB | Tight | Get file → | |
| Q8_0 | 8.0 GB | 9.9 GB | Too big | Get file → | |
| BF16 | 15.1 GB | 17.0 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_M (5.34 GB file, 7.2 GB memory). For 12 GB VRAM, Q5_K_M (5.76 GB file, 7.7 GB memory). For 16 GB VRAM, Q8_0 (8.03 GB file, 9.9 GB memory). BF16 requires 17 GB memory and is only for high‑end servers.
What GPU do you need for Cerberus E4B?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: the best quant that fits is Q5_K_M (5.8 GB file, about 7.7 GB in use).
- 12 GB VRAM: the best quant that fits is Q8_0 (8.0 GB file, about 9.9 GB in use).
- 16 GB VRAM: the best quant that fits is Q8_0 (8.0 GB file, about 9.9 GB in use).
- 24 GB VRAM: the best quant that fits is BF16 (15.1 GB file, about 17.0 GB in use).
Which Cerberus E4B quant should you choose?
Q4_K_M is the usual starting point for Cerberus E4B: it balances file size and quality well. Here it is a 5.3 GB file.
The Q8_0 file is 8.0 GB, about 1.5 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Cerberus E4B
How much VRAM does Cerberus E4B need?
About 7.2 GB at Q4_K_M with an 8K context window. That is the 5.3 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run Cerberus E4B on an 8 GB GPU?
Yes, with Q5_K_M: the 5.8 GB file needs about 7.7 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Cerberus E4B GGUF?
It depends on the quant: Q4_K_M is 5.3 GB, Q5_K_M is 5.8 GB, Q8_0 is 8.0 GB, BF16 is 15.1 GB.
What context length does Cerberus E4B support?
The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 1.4 GB at 8K and 5.6 GB at 32K.
Where can I download Cerberus E4B GGUF?
The files are on Hugging Face in the rrchat/Cerberus-E4B-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Cerberus E4B use?
Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run Cerberus E4B on 8 GB VRAM?
Yes, the Q4_K_M quantization uses 5.34 GB file size and requires about 7.2 GB memory at 8K context.
Does the model support audio input?
Audio is supported natively on the E4B variant, but only when using quantizations that include the frozen audio tower.
What license applies to this model?
The model is released under the Apache‑2.0 license, allowing free use and modification.