Model benchmarks
Qwen3.8-27B-Q4_K_S local LLM performance
As of September 2026, Qwen3.8-27B-Q4_K_S runs at up to 39.8 tok/s for local inference (best of 3 community benchmark runs across 1 GPU).
Model size
27B
Peak speed
39.8 tok/s
Average speed
36.1 tok/s
Avg PP
348.4 tok/s
Min memory
n/a
Max context
180,224 tokens
Avg output / run
12,607 tokens
Avg runtime / run
5m 49s
Avg quality
79.5
Benchmark runs
3
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 3 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 | 73.9 | 79.5 | 83.1 |
| Agent Workflow | 75.5 | 77.9 | 81.0 |
| Code Generation | 57.4 | 69.2 | 77.7 |
| Role Play & Narrative | 77.1 | 85.7 | 91.5 |
| Research & Analysis | 83.1 | 85.3 | 87.4 |
Performance by hardware and tool
Every hardware/tool/quantization combination Qwen3.8-27B-Q4_K_S has been benchmarked on, ranked by peak token generation speed. Last updated September 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| AMD Radeon RX 6700/6700 XT/6750 XT / 6800M/6850M XT | llama.cpp | Q4_K_S | 39.8 tok/s | 36.1 tok/s | n/a | 180,224 tokens | 79.5 | 3 |
Benchmark runs
All 3 Qwen3.8-27B-Q4_K_S runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is Qwen3.8-27B-Q4_K_S good for coding?
- In our benchmarks, Qwen3.8-27B-Q4_K_S scores 69.2/100 for coding. It runs at about 36.1 tok/s, so if you want more speed, local-qwen is faster (~60.2 tok/s) and still scores well for coding (84.8/100).
- Is Qwen3.8-27B-Q4_K_S good for agentic (tool-using) tasks?
- In our benchmarks, Qwen3.8-27B-Q4_K_S scores 77.9/100 for agentic workflows. It runs at about 36.1 tok/s, so if you want more speed, granite-4.2-8b-Q4_K_L is faster (~71.6 tok/s) and still scores well for agentic workflows (92.9/100).
- How fast is Qwen3.8-27B-Q4_K_S for local inference?
- Across 3 community benchmark runs, Qwen3.8-27B-Q4_K_S reaches up to 39.8 tok/s and averages 36.1 tok/s, with the fastest results on AMD Radeon RX 6700/6700 XT/6750 XT / 6800M/6850M XT.
- Which tools have been used to run Qwen3.8-27B-Q4_K_S?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.