SP

BeaverAI Spectre 37B v1a

SpectreGGUF by ReadyArtUpdated 2026-10-01
Download on Hugging Face
Parameters36.6B
Context256K
Layers72
Architecturegemma4
Smallest quant30.3 GB
Quick answer

BeaverAI Spectre 37B v1a needs about 69.5 GB of memory at Q6_K (a 30.3 GB file) with an 8K context window, which is more than a 48 GB GPU holds. Longer context needs more: at 32K tokens the KV cache alone adds about 154.6 GB.

BeaverAI Spectre 37B v1a – a 36.6B Gemma4 model in Q6_K format

BeaverAI Spectre 37B v1a is a 36.6B parameter Gemma4 model. It is a custom Q6_K quant of the Spectre-37B-v1a-GGUF model, with the Q8 embedding layer inserted. The quant adds about 500MB to size and VRAM usage.

Good at

  • Large context handling
  • High parameter count
  • Gemma4 architecture

Limits

  • Requires 69.5GB VRAM at 8k context
  • No license listed, assume Apache but confirm
  • No vision support

Best for

  • Research on large language models
  • Fine‑tuning with 8k context
  • Testing Gemma4 performance

Who it’s for. Researchers and developers needing a large Gemma4 model with a custom Q6_K quant. Suitable for those who can provide the necessary VRAM and want to experiment with the Q8 embedding layer.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q6_K30.3 GB 69.5 GB
Too big Get file →

The only available quant is Q6_K, file size 30.35GB. It requires approximately 69.5GB VRAM at 8k context. For lower VRAM budgets, a different quant would be needed, but none are listed.

What GPU do you need for BeaverAI Spectre 37B v1a?

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: no quant fits at an 8K context.
  • 32 GB VRAM: no quant fits at an 8K context.
  • 48 GB VRAM: no quant fits at an 8K context.

Which BeaverAI Spectre 37B v1a quant should you choose?

Q6_K (30.3 GB) is the middle option in this repository. Very close to the original quality for most uses, at a clearly larger size than 4-bit.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturegemma4
Parameters36.6B
Layers72
Trained context262,144 tokens
KV cache (fp16)4608 KB per token

Source

Quantized byReadyArt
FamilySpectre
Licensegemma
Downloads (HF)294
Last updated2026-10-01

Frequently asked questions about BeaverAI Spectre 37B v1a

How much VRAM does BeaverAI Spectre 37B v1a need?

About 69.5 GB at Q6_K with an 8K context window. That is the 30.3 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 BeaverAI Spectre 37B v1a on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 69.5 GB, so you would need a larger GPU or CPU offloading.

What is the file size of BeaverAI Spectre 37B v1a GGUF?

It depends on the quant: Q6_K is 30.3 GB.

What context length does BeaverAI Spectre 37B v1a support?

The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 38.7 GB at 8K and 154.6 GB at 32K.

Where can I download BeaverAI Spectre 37B v1a GGUF?

The files are on Hugging Face in the ReadyArt/BeaverAI_Spectre-37B-v1a repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does BeaverAI Spectre 37B v1a use?

Hugging Face lists the license as gemma. 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 this model on 8GB VRAM?

No. The Q6_K quant requires about 69.5GB VRAM at 8k context, far exceeding 8GB.

Is there a license for this model?

No license is listed; assume Apache but confirm with Drummer or BeaverAI.

Does this model support vision tasks?

No, it does not support vision.

Similar models

Browse all →

More from ReadyArt

All by ReadyArt →

More Spectre models

All Spectre →

Other large models

All large models →