Qwen3.8-27B TTURBO 709 Uncensored - BF16 GGUF

APEX Quantization Series by IsValorum

[!IMPORTANT] Independently computed using dedicated cloud compute clusters, funded 100% out-of-pocket.

If you consider this release useful, consider supporting on Ko-fi. Every contribution directly sustains future cloud compute hours and helps support my university studies.

[!NOTE] MTP included: the 15 tensors in blk.64.* are part of this BF16 GGUF. There is no separate draft file.

Base model

• DavidAU/Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored (Apache-2.0)
• This is the BF16 reference for the DAU TTURBO 709 Uncensored release, converted from the original safetensors with llama.cpp master.
• The DAU chat template is embedded in the GGUF and is also provided as chat_template.jinja.

Files

The vision projector is the original mmproj-BF16.gguf from the base GGUF repository.

SHA256

d2932275ad1a4316f482967b1cde41a5e5461b12965b4f558ba91eab1095c36f

Usage

llama-cli \
  -m Qwen3.8-27B-TTURBO-709-Uncen-BF16.gguf \
  -p "Explain why the sky is blue in one paragraph." \
  -ngl 99 -c 32768 -t 8

llama-server \
  -m Qwen3.8-27B-TTURBO-709-Uncen-BF16.gguf \
  --mmproj mmproj-BF16.gguf \
  --jinja -ngl 99 -c 32768

Use this BF16 file as the reference. The MiniPlus and NanoPlus V3 tiers are the deployment builds.

Limitations

• Raw BF16 reference, not a quantized tier.
• Base model benchmarks are self-reported by DavidAU.
• Vision and MTP were not independently evaluated in this archive.

License

Apache-2.0, inherited from the base model. See the base model card for the original terms.