Model benchmarks
Qwen3.8-27B-Q8_0 local LLM performance
As of September 2026, Qwen3.8-27B-Q8_0 runs at up to 44.8 tok/s for local inference (best of 3 community benchmark runs across 1 GPU).
Model size
27B
Peak speed
44.8 tok/s
Average speed
40.6 tok/s
Avg PP
465.4 tok/s
Min memory
26.4 GB
Max context
65.536 tokens
Avg output / run
35.573 tokens
Avg runtime / run
14m 49s
Avg quality
85.1
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 | 82.2 | 85.1 | 88.1 |
| Agent Workflow | 85.2 | 86.8 | 89.1 |
| Code Generation | 68.6 | 77.0 | 83.7 |
| Role Play & Narrative | 89.3 | 90.7 | 91.8 |
| Research & Analysis | 82.5 | 86.0 | 91.5 |
Performance by hardware and tool
Every hardware/tool/quantization combination Qwen3.8-27B-Q8_0 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 7900 XT/7900 XTX/7900M | llama.cpp | Q8_0 | 44.8 tok/s | 40.6 tok/s | 26.4 GB | 65.536 tokens | 85.1 | 3 |
Benchmark runs
All 3 Qwen3.8-27B-Q8_0 runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is Qwen3.8-27B-Q8_0 good for coding?
- In our benchmarks, Qwen3.8-27B-Q8_0 scores 77.0/100 for coding. It runs at about 40.6 tok/s, so if you want more speed, unsloth/Qwen3.8-27B-GGUF:IQ3_S is faster (~104.3 tok/s) and still scores well for coding (83.0/100).
- Is Qwen3.8-27B-Q8_0 good for agentic (tool-using) tasks?
- In our benchmarks, Qwen3.8-27B-Q8_0 scores 86.8/100 for agentic workflows. It runs at about 40.6 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-Q8_0 for local inference?
- Across 3 community benchmark runs, Qwen3.8-27B-Q8_0 reaches up to 44.8 tok/s and averages 40.6 tok/s, with the fastest results on AMD Radeon RX 7900 XT/7900 XTX/7900M.
- How much memory does Qwen3.8-27B-Q8_0 need?
- The leanest observed configuration used about 26.4 GB of memory (quantizations tested: Q8_0).
- Which tools have been used to run Qwen3.8-27B-Q8_0?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.