MU

Homura 30B

MuseGGUF by hyrelabsUpdated 2026-08-22
Download on Hugging Face
Parameters28B
Context128K
Layers52
Architecturemuse-glimmer
Smallest quant16.9 GB
Quick answer

Homura 30B needs about 17.9 GB of memory at Q4_K_M (a 16.9 GB file) with an 8K context window, so it fits on a 24 GB GPU. Longer context needs more: at 32K tokens the KV cache alone adds about 1.7 GB.

Homura 30B – a 28‑billion‑parameter agent‑tuned model for local tool‑calling

Homura 30B is a 28‑billion‑parameter model derived from Muse Glimmer 30B. It is fine‑tuned for autonomous agent tasks, with reliable tool use and multi‑step reasoning. The model runs locally via llama.cpp and is available in Q4_K_M quantization.

Good at

  • Reliable tool calling
  • Multi‑step reasoning
  • Local deployment on consumer hardware

Limits

  • No vision encoder in the quantized file
  • Requires llama.cpp ≥ b10353
  • Large context size (131072 tokens) may need more VRAM

Best for

  • Building autonomous agents
  • Running local LLM workflows
  • Tool‑integrated code generation

Who it’s for. Developers and researchers who need a locally deployable agent model with robust tool use and long‑horizon reasoning. The model is suitable for projects that require no cloud access and can run on consumer hardware.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q4_K_M16.9 GB 17.9 GB
Too big Get file →

The Q4_K_M file is 16.94 GB and requires about 17.9 GB of VRAM at 8 k context. For 8 GB VRAM, a higher‑quantization such as Q8_0 would be needed, but it is not provided. For 16 GB VRAM, Q4_K_M is suitable. For 32 GB VRAM, any quant is fine.

What GPU do you need for Homura 30B?

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: no quant fits at an 8K context.
  • 12 GB VRAM: no quant fits at an 8K context.
  • 16 GB VRAM: no quant fits at an 8K context.
  • 24 GB VRAM: the best quant that fits is Q4_K_M (16.9 GB file, about 17.9 GB in use).

Which Homura 30B quant should you choose?

Q4_K_M is the usual starting point for Homura 30B: it balances file size and quality well. Here it is a 16.9 GB file.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturemuse-glimmer
Parameters28B
Layers52
Trained context131,072 tokens
KV cache (fp16)52 KB per token

Source

Quantized byhyrelabs
FamilyMuse
Licenseapache-2.0
Downloads (HF)82,528
Last updated2026-08-22

Frequently asked questions about Homura 30B

How much VRAM does Homura 30B need?

About 17.9 GB at Q4_K_M with an 8K context window. That is the 16.9 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 Homura 30B on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 17.9 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Homura 30B GGUF?

It depends on the quant: Q4_K_M is 16.9 GB.

What context length does Homura 30B support?

The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 0.4 GB at 8K and 1.7 GB at 32K.

Where can I download Homura 30B GGUF?

The files are on Hugging Face in the hyrelabs/Homura-30B-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 Homura 30B 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 Homura 30B on 8 GB VRAM?

The Q4_K_M quantization requires about 17.9 GB VRAM at 8 k context, so 8 GB VRAM is insufficient.

Does Homura 30B support vision?

The quantized file contains only the language tower; the vision encoder is not included.

What temperature should I use for tool calling?

Use temperature ≤ 0.3 for reliable tool calling; higher temperatures reduce accuracy.

Other GGUFs of Muse-Glimmer-30B-heretic

Similar models

Browse all →

More from hyrelabs

All by hyrelabs →

More Muse models

All Muse →

Other medium models

All medium models →