BT

BTL 4 Compact

BtlGGUF by badtheorylabsUpdated 2026-08-07
Download on Hugging Face
Parameters34.7B
Context256K
Layers40
Architectureqwen35moe
Smallest quant10.0 GB
Quick answer

BTL 4 Compact needs about 11.1 GB of memory at IQ2_XXS (a 10.0 GB file) with an 8K context window, so it fits on a 12 GB GPU. Longer context needs more: at 32K tokens the KV cache alone adds about 2.7 GB.

BTL 4 Compact – a 35 B mixture‑of‑experts model in a single 9.97 GB GGUF file

BTL 4 Compact is a 35 B Qwen‑3.5 Moe model fine‑tuned for tool use, software engineering and long‑horizon agent work. It retains 94.1 % of the full‑precision model’s behaviour in a single 9.97 GB file. The model is built for reasoning before answering and is verbose by default.

Good at

  • Tool calling – 73.5 % BFCL v4 AST
  • Competitive programming – 99.1 % easy / 86.7 % medium on LiveCodeBench
  • Long context – 262 K native tokens

Limits

  • Requires llama.cpp with qwen3_5_moe support
  • Needs --jinja and --reasoning-format deepseek flags
  • Token‑hungry on hard problems – budget accordingly

Best for

  • Running an agent that calls tools
  • Solving competitive programming problems
  • Processing long documents with 262 K context

Who it’s for. Developers and researchers who need a large‑scale reasoning model that can be run on existing hardware. The model is ideal for building agents that perform code generation, tool use, and long‑horizon tasks without requiring a separate base download.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ2_XXS10.0 GB 11.1 GB
Too big Get file →

The only available quant is IQ2_XXS, 9.97 GB. It requires approximately 11.1 GB of VRAM at an 8 K context. For systems with less VRAM, a higher‑bit quant would be needed, but none are provided in the file list.

What GPU do you need for BTL 4 Compact?

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: the best quant that fits is IQ2_XXS (10.0 GB file, about 11.1 GB in use).

Which BTL 4 Compact quant should you choose?

IQ2_XXS (10.0 GB) is the middle option in this repository. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architectureqwen35moe
Parameters34.7B (approximate, mixture of experts)
Layers40
Trained context262,144 tokens
KV cache (fp16)80 KB per token

Source

Quantized bybadtheorylabs
FamilyBtl
Licenseapache-2.0
Downloads (HF)82,646
Last updated2026-08-07

Frequently asked questions about BTL 4 Compact

How much VRAM does BTL 4 Compact need?

About 11.1 GB at IQ2_XXS with an 8K context window. That is the 10.0 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 BTL 4 Compact on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 11.1 GB, so you would need a larger GPU or CPU offloading.

What is the file size of BTL 4 Compact GGUF?

It depends on the quant: IQ2_XXS is 10.0 GB.

What context length does BTL 4 Compact support?

The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 0.7 GB at 8K and 2.7 GB at 32K.

Where can I download BTL 4 Compact GGUF?

The files are on Hugging Face in the badtheorylabs/BTL-4-Compact repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does BTL 4 Compact 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 BTL 4 Compact on 8 GB VRAM?

No. The IQ2_XXS quant requires about 11.1 GB of VRAM at an 8 K context, which exceeds 8 GB.

Does BTL 4 Compact support vision?

No. The model does not have vision capabilities.

What flags are required to run the model?

Use --jinja and --reasoning-format deepseek; do not override the chat template.

Other GGUFs of BTL-4

Similar models

Browse all →

More from badtheorylabs

All by badtheorylabs →

More Btl models

All Btl →

Other large models

All large models →