---

license: other

license_name: polyform-small-business-1.0.0-bottlecap-personal-noncommercial

license_link: https://huggingface.co/ngquocvinh/ThinkingCap-Qwen3.8-27B-GGUF/blob/main/LICENSE

base_model: bottlecapai/ThinkingCap-Qwen3.8-27B

base_model_relation: quantized

library_name: llama.cpp

pipeline_tag: text-generation

tags:

• gguf
• llama.cpp
• thinkingcap
• quantized

---

ThinkingCap-Qwen3.8-27B GGUF

Community GGUF quantizations of bottlecapai/ThinkingCap-Qwen3.8-27B.

<div align="center" style="background-color:#f59e0b;color:#ffffff;padding:16px 20px;border-radius:10px;line-height:1.7;">

☕ If this GGUF made your day easier, a coffee would make mine.<br>

<a href="https://ko-fi.com/ngquocvinh" style="color:#ffffff;"><strong style="color:#ffffff;">Send a coffee ☕</strong></a><br>

I build and test these releases myself. Your coffee helps keep me going.<br>

Thank you for supporting this work.

</div>

About ThinkingCap-Qwen3.8-27B

This is a community GGUF package of BottleCap AI's 27B dense vision-language model, based on Qwen3.8-27B. The upstream model card describes hybrid linear and full attention, image and video understanding, an MTP head, and a native context length of 262,144 tokens. See the upstream model card for its capabilities, evaluation results, and recommended sampling settings. Those upstream task measurements apply to the upstream checkpoint and its exact quantized files; this package does not claim equivalent scores for these files.

The package includes the upstream F16 multimodal projector. The ten quantizations from the initial release passed text generation smoke tests. Nine additional formats have also passed the same check; all 19 quantizations are published here. A held-out fidelity comparison to the locked F16 source has not been completed, so this card makes no fidelity ranking or quality percentage claim. Image input did not pass smoke validation with the pinned llama.cpp build: image encoding completed, but token decoding failed on both CUDA and CPU profiles. Treat image input as unverified with this runtime/package combination.

No training or fine-tuning was performed for this release. These are community GGUF quantizations, not an official BottleCap AI release or endorsement.

The Q2-and-lower formats, including IQ2 and IQ1, are experimental, low-memory options; their task behavior has not been evaluated in this release.

Quick start

The example below uses Q4_K_M and the chat template embedded in the GGUF. Replace the filename with another quantization from this repository's file tree if needed. The release validates text generation; image input remains unverified with the pinned llama.cpp build.

./llama-cli \
  -m ThinkingCap-Qwen3.8-27B-Q4_K_M.gguf \
  --jinja \
  --reasoning-effort xhigh \
  --temp 1.0 --top-p 0.95 --top-k 20 --min-p 0.0 \
  -p 'Answer briefly in English: What makes a good incident report?' \
  -n 256 -c 4096 -ngl auto

Reproducibility and validation

The quantized GGUF files are built directly from the locked upstream F16 GGUF. A ThinkingCap-specific importance matrix is used for the lower-bit formats; Q8_0 is built without it. The image projector remains F16, but image generation was not validated successfully with the pinned runtime. All 19 published quantizations passed text load-and-generate smoke tests. No held-out fidelity evaluation or VRAM-fit measurement has been completed, so this card makes no fidelity ranking or precise VRAM claim.

• SHA256 checksums
• Reproducibility manifest
• Artifact measurements
• Calibration data notice

The manifest records upstream revisions, input hashes, the runtime commit, calibration inputs, quantization commands, and validation profile. The included calibration texts and importance matrix support reproduction. Raw conversion, build, quantization, and smoke-test logs are kept outside this public repository.

License and attribution

The ThinkingCap materials are distributed under the PolyForm Small Business License 1.0.0, with BottleCap AI's additional individual personal, non-commercial permission in LICENSE. The complete Qwen Apache-2.0 notice is included in NOTICE and LICENSE-Apache-2.0-Qwen.txt. Review the license terms before use, especially for organizational or commercial deployment.

These are community GGUF quantizations, not an official BottleCap AI release or endorsement.