---

base_model:

• Qwen/Qwen3.8-27B

base_model_relation: finetune

license: apache-2.0

license_link: LICENSE

library_name: gguf

pipeline_tag: image-text-to-text

tags:

- token-efficient

- efficient-thinking

- qwen3.8

- qwen3.8-27b

- terse

- agentionai

---

<a href="https://www.agention.ai/"><img src="signal27b.png" alt="AgentionAI — Signal 3.8 27B" width="100%"></a>

Signal 3.8 27B

This is Qwen3.8-27B that gets to the answer faster.

AgentionAI Signal is a minimally invasive fine-tune of Qwen3.8-27B designed for lower generation latency and better token efficiency. On our held-out general-prompt evaluation, Signal produces 57% fewer answer tokens and uses 52% fewer thinking tokens, while matching or improving the measured answer quality of the base model.

The result is substantially faster end-to-end generation: on typical chat prompts, Signal can finish in less than half the wall time of the untouched Qwen3.8-27B on the same hardware.

Signal gets there by being more direct rather than by truncating answers. It removes unnecessary preambles, excessive formatting, sign-offs, and explanatory narration while preserving the substance of the response. In thinking mode, it keeps the useful reasoning steps while spending fewer tokens describing the process.

Signal is trained by self-distillation: on Qwen3.8-27B's own answers, generated under an instruction to be direct that the released model no longer needs. No external data and no other model's outputs went into it, which is why it keeps the base model's knowledge and voice intact.

It is a drop-in GGUF replacement for llama.cpp setups that already run Qwen3.8-27B.

What changes, measured

We evaluated Signal against the untouched Qwen3.8-27B Q8_0 using the same server, sampling settings, prompts, and otherwise identical model file. All prompts in these evaluations were held out from tuning.

| | base Q8_0 | Signal | change |

|---|---|---|---|

| general answers, median tokens | 243 | 104 | -57% |

| answers opening with a preamble ("Sure!", "Great question") | 13% | 0% | gone |

| answers with markdown headers | 47% | 18% | -62% |

| answers with bold | 85% | 52% | -39% |

| coding answers, median tokens | 159 | 142 | -11% |

| coding answers, p90 tokens | 1026 | 914 | -11% |

Thinking mode, same prompts with reasoning on:

| | base Q8_0 | Signal | change |

|---|---|---|---|

| reasoning tokens, general prompts, median | 153 | 74 | -52% |

| reasoning tokens, coding prompts, median | 225 | 166 | -26% |

| reasoning tokens, GSM8K, median | 119 | 81 | -32% |

Quality, exact match on GSM8K:

| | base Q8_0 | Signal |

|---|---|---|

| thinking off, 60 problems | 98.3% | 98.3% |

| thinking on, 40 problems | 92.5% | 95.0% |

Shorter is not cheaper: no answer in the 100-prompt style set was cut off early (0 answers

ending on a header or a colon, 0 unclosed code blocks), and no reasoning trace in 50

thinking-mode outputs looped or hit the token cap.

Faster with speculative decoding

Qwen3.8-27B carries a built-in multi-token-prediction draft head. Signal's answers are more predictable, so the drafter agrees with the model more often:

Draft acceptance and decode speed with --spec-type draft-mtp, both models Q8_0 on the same

machine (Strix Halo, Vulkan), 200-token greedy runs for the fixed-draft rows:

| prompt / draft length | base acceptance | Signal acceptance | decode speed vs base |

|---|---:|---:|---:|

| prose, draft 3 | 39% | 47% | +10% |

| prose, draft 4 | 35% | 28% | -9% |

| structured output (JSON), draft 3 | 72% | 94% | +20% |

| structured output (JSON), draft 4 | 66% | 87% | +22% |

| chat prompts, sampled at 0.7, adaptive draft ≤4 (40 prompts) | 57% | 60% | — |

Combined with the shorter answers, a typical chat reply finishes in well under **half the wall

time** of the base model on the same hardware. Because of the specific finetuning there is no degredation over long context length.

Files

Six tiers from IQ4_XS to Q8_0, one folder each. AP = Agention Precision: tier

names refer to the base ftype, the per-tensor types are chosen for accuracy per gigabyte using Agention Precisionrecipes. All built from Signal's BF16 using our custom imatrix. VRAM is the file size: this is a dense model,

nothing offloads.

| tier | size | eff. bpw | KLD held-out | top-1 | KLD wikitext | what |

|---|---:|---:|---:|---:|---:|---|

| Q8_0 | 27.05 GiB | 8.26 | 0.0040 | 95.2% | 0.0045 | reference quality; every number below was measured on this file |

| AP-Q6_K | 20.89 GiB | 6.57 | 0.0055 | 95.0% | 0.0075 | Q6_K with imatrix |

| AP-Q5_K_M | 18.19 GiB | 5.72 | 0.0081 | 94.4% | 0.0094 | Q5_K_M with imatrix |

| AP-Q4_K_XL | 16.35 GiB | 5.14 | 0.0124 | 93.9% | 0.0148 | precision tier — matched unsloth's UD-Q4_K_XL on the base model at the same size |

| AP-Q4_K_M | 15.83 GiB | 4.98 | 0.0184 | 92.6% | 0.0218 | fast tier — +20% prefill over UD-Q4_K_M on the base model at the same size, some quality traded |

| AP-IQ4_XS | 13.27 GiB | 4.17 | 0.0287 | 91.1% | 0.0327 | UD-IQ4_XS per-tensor map with our imatrix — same size |

Which one: AP-Q4_K_XL if 16.5 GiB fits, AP-IQ4_XS for the 13 GiB slot, AP-Q4_K_M when prefill

speed matters more than the last bit of quality, AP-Q6_K or Q8_0 when memory is no object. KLD and top-1 agreement are measured against Signal's own BF16 (-c 2048, 60 chunks, the

campaign's frozen held-out and wikitext-2 corpora). Every campaign recipe lands within 2% of the

KLD it scored on the base model, so the tune costs nothing at any tier.

Running

Thinking on and off both work; the chat template is the original Qwen3.8 template.

Sampling: temperature 0.7, top-p 0.95, top-k 20, min-p 0, as in the commands below. Use sampling

rather than greedy decoding. We saw a single loop at temperature:0.

<details open>

<summary>llama.cpp</summary>

llama-server -hf agentionai/Signal-3.8-27B-GGUF:AP-Q4_K_XL \
  --jinja -ngl 999 -fa on -c 32768 \
  --temp 0.7 --top-p 0.95 --top-k 20 --min-p 0

Add the built-in draft head for the throughput above (needs a build with --spec-type draft-mtp):

  --spec-type draft-mtp --spec-draft-n-max 4

Use --spec-draft-adaptive if running halo-box

Thinking is on by default. To turn it off per request, send

"chat_template_kwargs": {"enable_thinking": false} with the chat completion.

</details>

<details>

<summary>Ollama / LM Studio</summary>

Import the GGUF as any Qwen3.8-27B file. Use the sampling settings above; the template is

embedded in the file.

</details>

Vision

Signal keeps Qwen3.8-27B's image input. The vision encoder and projector are untouched by the

tune (it changes the text output layer only), so the base model's own mmproj-BF16.gguf is at

the repository root, 0.87 GiB. Download it alongside any tier:

Support AgentionAI

Signal3.8 is released freely. If it saves you compute or makes Qwen more useful, you can sponsor continued tuning, quantization and benchmarking on GitHub.