Model benchmarks
Qwen3.8-4B-Q5_K_M local LLM performance
As of October 2026, Qwen3.8-4B-Q5_K_M runs at up to 55.1 tok/s for local inference (best of 5 community benchmark runs across 1 GPU).
Model size
4B
Peak speed
55.1 tok/s
Average speed
52.7 tok/s
Avg PP
871.2 tok/s
Min memory
4.4 GB
Max context
32,768 tokens
Avg output / run
11,725 tokens
Avg runtime / run
3m 57s
Avg quality
65.5
Benchmark runs
5
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 5 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 | 58.5 | 65.5 | 69.9 |
| Agent Workflow | 34.1 | 64.7 | 84.4 |
| Code Generation | 41.3 | 46.1 | 49.9 |
| Role Play & Narrative | 71.4 | 77.6 | 82.7 |
| Research & Analysis | 70.3 | 73.7 | 78.0 |
Performance by hardware and tool
Every hardware/tool/quantization combination Qwen3.8-4B-Q5_K_M has been benchmarked on, ranked by peak token generation speed. Last updated October 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| NVIDIA GeForce RTX 4060 Laptop GPU | llama.cpp | Q5_K_M | 55.1 tok/s | 52.7 tok/s | 4.4 GB | 32,768 tokens | 65.5 | 5 |
Benchmark runs
All 5 Qwen3.8-4B-Q5_K_M runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is Qwen3.8-4B-Q5_K_M good for coding?
- In our benchmarks, Qwen3.8-4B-Q5_K_M scores 46.1/100 for coding. It runs at about 52.7 tok/s, so if you want more speed, nvidia/Qwen3.8-27B-NVFP4 is faster (~63.3 tok/s) and still scores well for coding (82.7/100).
- Is Qwen3.8-4B-Q5_K_M good for agentic (tool-using) tasks?
- In our benchmarks, Qwen3.8-4B-Q5_K_M scores 64.7/100 for agentic workflows. It runs at about 52.7 tok/s, so if you want more speed, Nex-N2.5-mini-APEX-Mini is faster (~88.2 tok/s) and still scores well for agentic workflows (91.3/100).
- How fast is Qwen3.8-4B-Q5_K_M for local inference?
- Across 5 community benchmark runs, Qwen3.8-4B-Q5_K_M reaches up to 55.1 tok/s and averages 52.7 tok/s, with the fastest results on NVIDIA GeForce RTX 4060 Laptop GPU.
- How much memory does Qwen3.8-4B-Q5_K_M need?
- The leanest observed configuration used about 4.4 GB of memory (quantizations tested: Q5_K_M).
- Which tools have been used to run Qwen3.8-4B-Q5_K_M?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.