This is a vanilla quantization of Qwen/Qwen3.8-27B. It is not a fine-tune, merge, ablation, alignment change, or chat-template modification. The source weights are pinned to commit 1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0.

The official checkpoint uses Qwen3_5ForConditionalGeneration / qwen3_5 as its internal architecture identifier. That string does not mean these weights came from a Qwen3.5 model.

Conversion

json

{
  "algorithm": "llama.cpp stable K-quants and IQ quants from one F16 GGUF",
  "bit_width": [
    2,
    3,
    4,
    5,
    6,
    8
  ],
  "group_size": "format_defined",
  "calibration_source": "none for K-quants; local representative prompts for IQ variants if required"
}

Source tensor inventory: 1199 tensors, including 333 vision tensors and 15 source MTP tensors.
Conversion tool/runtime requirement: llama.cpp / 5f754ea0e2fd21e1213db7ebebfd65d938d9d69c.
Artifact size: 206.295 GB (decimal).
Expected hardware: Apple Silicon Metal or CPU; 120 GB temporary headroom.

Calibration source: none for K-quants; local representative prompts for IQ variants if required.

Component status

Text: passed release tests.
Vision/video: passed deterministic local image tests.
Tool calling: passed all native XML tool tests.
MTP: source MTP tensors were retained by the structural gate; this repository does not claim speculative acceleration.
Chat template, tokenizer, processor, generation config, and special-token IDs: checked against the locked source by the structural gate.
Quality comparison: passed against the locked BF16 source using the exact same functional cases. Semantic similarity uses sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2 at e8f8c211226b894fcb81acc59f3b34ba3efd5f42 as a measured proxy, not as ground-truth accuracy.
Longest recorded validation prompt: 73 prompt tokens. This is a measured test boundary, not a claim that the architectural maximum was exercised.

Validation results

json

{
  "release_gate": "PASS",
  "text": [
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true,
    true
  ],
  "tools": [
    true,
    true,
    true,
    true,
    true
  ],
  "vision": [
    true,
    true,
    true
  ],
  "mtp": {
    "passed": true,
    "acceleration_claimed": false,
    "retention_gate": "GGUF tensor and nextn metadata inspection",
    "advertise_acceleration": false
  },
  "bf16_source_comparison": {
    "passed": true,
    "mean_semantic_similarity": 0.9082151889801026,
    "exact_matches": 5,
    "measurements": {
      "average_generation_tps": 8.600119274947904,
      "peak_memory_gb": null,
      "artifact_bytes": 206294771716,
      "maximum_prompt_tokens_tested": 73,
      "loop_rate": 0.0
    },
    "evaluator": {
      "repo_id": "sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2",
      "revision": "e8f8c211226b894fcb81acc59f3b34ba3efd5f42",
      "pooling": "attention-mask mean pooling followed by L2 normalization",
      "maximum_tokens": 256
    }
  },
  "bf16_fixed_logit_comparison": "not applicable to the original 30-slot matrix"
}

No acceleration is advertised unless the MTP report contains a measured throughput improvement. Exact measurements are artifact-, prompt-, context-, and hardware-specific.

Inference

bash

git clone https://github.com/ggml-org/llama.cpp && cd llama.cpp
git checkout 5f754ea0e2fd21e1213db7ebebfd65d938d9d69c
cmake -S . -B build -DGGML_METAL=ON -DCMAKE_BUILD_TYPE=Release
cmake --build build --config Release -j
hf download Chungulus/Qwen3.8-27B-GGUF --local-dir ../qwen38-gguf
cd ../qwen38-gguf
../llama.cpp/build/bin/llama-mtmd-cli -m ./Qwen3.8-27B-Q4_K_M.gguf --mmproj ./mmproj-Qwen3.8-27B-F16.gguf -p 'Describe the image.' --image ./image.png

Use the exact source chat-template controls for thinking (enable_thinking, reasoning_effort, and preserve_thinking) and the native Qwen tool format.

Limitations

Quantization can reduce quality, especially at very low bit widths. Runtime support for the hybrid Gated DeltaNet/full-attention graph, vision tower, projector, processor, and MTP component is format-specific. A loader that reads only a language tensor is not sufficient. Tested context length and resource measurements are recorded in validation_result.json; untested context lengths must not be inferred from the architectural maximum.

License and attribution

The parent model and this unmodified quantization are distributed under the source model's Apache-2.0 license. See the official Qwen3.8-27B repository for the upstream model card and attribution.