Is unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M good for coding?

In our benchmarks, unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M scores 68.8/100 for coding. It runs at about 61.8 tok/s, so if you want more speed, ornith-ai/Ornith-1.5-35B-A3B-GGUF is faster (~80.3 tok/s) and still scores well for coding (82.1/100).

Is unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M good for agentic (tool-using) tasks?

In our benchmarks, unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M scores 72.4/100 for agentic workflows. It runs at about 61.8 tok/s, so if you want more speed, swift-iq2_xs is faster (~91.5 tok/s) and still scores well for agentic workflows (91.4/100).

How fast is unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M for local inference?

Across 12 community benchmark runs, unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M reaches up to 121.5 tok/s and averages 61.8 tok/s, with the fastest results on Apple M5 Ultra.

How much memory does unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M need?

The leanest observed configuration used about 19.8 GB of memory (quantizations tested: Q4_K_M).

Which tools have been used to run unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M?

Benchmarks were submitted using Splash, llama.cpp, splash, vLLM. Results are community-contributed and updated as new runs arrive.

Model benchmarks

unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M local LLM performance

As of November 2026, unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M runs at up to 121.5 tok/s for local inference (best of 12 community benchmark runs across 6 GPUs).

Splashllama.cppsplashvLLMQ4_K_M

ShareReddit X

Model size

27B

Peak speed

121.5 tok/s

Average speed

61.8 tok/s

Avg PP

516.4 tok/s

Min memory

19.8 GB

Max context

262,144 tokens

Avg output / run

28,494 tokens

Avg runtime / run

8m 53s

Avg quality

77.6

Benchmark runs

GPUs tested

Quality by task

Average LLM-judged quality (0–100) with the run-to-run spread shown as a P5–P95 band, overall and for each benchmark task, across all 12 runs. The low and high columns show how much the judge’s score varies between runs, and need at least two runs to display.

Task	P5 (low)	Avg	P95 (high)
Overall	55.5	77.6	87.8
Agent Workflow	35.0	72.4	91.5
Code Generation	21.2	68.8	86.4
Role Play & Narrative	74.1	86.2	94.4
Research & Analysis	77.3	82.9	88.6

Performance by hardware and tool

Every hardware/tool/quantization combination unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M has been benchmarked on, ranked by peak token generation speed. Last updated November 2026.

Hardware	Tool	Quant	Best tok/s	Avg tok/s	Memory	Context	Quality	Runs
Apple M5 Ultra	splash	Q4_K_M	121.5 tok/s	114.5 tok/s	21.4 GB	65,536 tokens	76.8	2
Apple M5 Max	splash	Q4_K_M	86.3 tok/s	83.0 tok/s	20.9 GB	65,536 tokens	83.1	2
Apple M5 Max	vLLM	Q4_K_M	65.6 tok/s	65.6 tok/s	n/a	8,192 tokens	46.6	1
AMD Radeon AI PRO R9700	llama.cpp	Q4_K_M	55.6 tok/s	55.6 tok/s	21.3 GB	65,536 tokens	86.2	1
NVIDIA GeForce GTX 1070 with Max-Q Design	llama.cpp	Q4_K_M	44.8 tok/s	44.8 tok/s	n/a	65,536 tokens	81.8	1
Apple M4 Pro	Splash	Q4_K_M	38.6 tok/s	38.0 tok/s	19.8 GB	262,144 tokens	77.6	4
Apple M5 Pro	splash	Q4_K_M	28.3 tok/s	28.3 tok/s	22.0 GB	65,536 tokens	86.2	1

Benchmark runs

All 12 unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Open run page →

Frequently asked questions

Is unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M good for coding?: In our benchmarks, unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M scores 68.8/100 for coding. It runs at about 61.8 tok/s, so if you want more speed, ornith-ai/Ornith-1.5-35B-A3B-GGUF is faster (~80.3 tok/s) and still scores well for coding (82.1/100).
Is unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M good for agentic (tool-using) tasks?: In our benchmarks, unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M scores 72.4/100 for agentic workflows. It runs at about 61.8 tok/s, so if you want more speed, swift-iq2_xs is faster (~91.5 tok/s) and still scores well for agentic workflows (91.4/100).
How fast is unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M for local inference?: Across 12 community benchmark runs, unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M reaches up to 121.5 tok/s and averages 61.8 tok/s, with the fastest results on Apple M5 Ultra.
How much memory does unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M need?: The leanest observed configuration used about 19.8 GB of memory (quantizations tested: Q4_K_M).
Which tools have been used to run unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M?: Benchmarks were submitted using Splash, llama.cpp, splash, vLLM. Results are community-contributed and updated as new runs arrive.