Model benchmarks
unsloth/Qwen3.8-27B-GGUF:Q4_K_M local LLM performance
As of August 2026, unsloth/Qwen3.8-27B-GGUF:Q4_K_M runs at up to 90.3 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
27B
Peak speed
90.3 tok/s
Average speed
89.2 tok/s
Min memory
13.2 GB
Max context
131,072 tokens
Avg output / run
61,533 tokens
Avg runtime / run
10m 31s
Avg quality
85.8
Benchmark runs
2
GPUs tested
1
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 2 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 | 85.4 | 85.8 | 86.1 |
| Agent Workflow | 86.3 | 86.5 | 86.7 |
| Code Generation | 83.4 | 83.7 | 84.0 |
| Role Play & Narrative | 86.4 | 88.0 | 89.5 |
| Research & Analysis | 84.4 | 84.9 | 85.3 |
Performance by hardware and tool
Every hardware/tool/quantization combination unsloth/Qwen3.8-27B-GGUF:Q4_K_M has been benchmarked on, ranked by peak token generation speed. Last updated August 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| NVIDIA GeForce RTX 4090 | llama.cpp | Q4_K_M | 90.3 tok/s | 89.2 tok/s | 13.2 GB | 131,072 tokens | 85.7 | 2 |
Frequently asked questions
- Is unsloth/Qwen3.8-27B-GGUF:Q4_K_M good for coding?
- In our benchmarks, unsloth/Qwen3.8-27B-GGUF:Q4_K_M scores 83.7/100 for coding, among the top 2 for coding on consumer hardware (≤24 GB VRAM). It runs at about 89.2 tok/s, so if you want more speed, Tiel-Coder-35B-A3B-MLX-oQ4e-MTP is faster (~123.6 tok/s) and still scores well for coding (76.5/100).
- Is unsloth/Qwen3.8-27B-GGUF:Q4_K_M good for agentic (tool-using) tasks?
- In our benchmarks, unsloth/Qwen3.8-27B-GGUF:Q4_K_M scores 86.5/100 for agentic workflows. It runs at about 89.2 tok/s, so if you want more speed, openai/gpt-oss-20b:2 is faster (~154.8 tok/s) and still scores well for agentic workflows (87.2/100).
- How fast is unsloth/Qwen3.8-27B-GGUF:Q4_K_M for local inference?
- Across 2 community benchmark runs, unsloth/Qwen3.8-27B-GGUF:Q4_K_M reaches up to 90.3 tok/s and averages 89.2 tok/s, with the fastest results on NVIDIA GeForce RTX 4090.
- How much memory does unsloth/Qwen3.8-27B-GGUF:Q4_K_M need?
- The leanest observed configuration used about 13.2 GB of memory (quantizations tested: Q4_K_M).
- Which tools have been used to run unsloth/Qwen3.8-27B-GGUF:Q4_K_M?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.