Model benchmarks
qwen3.8-9b-distill-mlx@4bit local LLM performance
As of November 2026, qwen3.8-9b-distill-mlx@4bit runs at up to 42.2 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
9B
Peak speed
42.2 tok/s
Average speed
41.7 tok/s
Avg PP
247.1 tok/s
Min memory
7.4 GB
Max context
65,536 tokens
Avg output / run
13,531 tokens
Avg runtime / run
5m 51s
Avg quality
65.9
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 | 64.3 | 65.9 | 67.5 |
| Agent Workflow | 78.4 | 83.5 | 88.5 |
| Code Generation | 34.2 | 43.2 | 52.2 |
| Role Play & Narrative | 67.0 | 74.2 | 81.4 |
| Research & Analysis | 58.1 | 62.7 | 67.4 |
Performance by hardware and tool
Every hardware/tool/quantization combination qwen3.8-9b-distill-mlx@4bit 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 M1 Max | LM Studio | 4bit | 42.2 tok/s | 41.7 tok/s | 7.4 GB | 65,536 tokens | 65.9 | 2 |
Benchmark runs
All 2 qwen3.8-9b-distill-mlx@4bit runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is qwen3.8-9b-distill-mlx@4bit good for coding?
- In our benchmarks, qwen3.8-9b-distill-mlx@4bit scores 43.2/100 for coding. It runs at about 41.7 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 qwen3.8-9b-distill-mlx@4bit good for agentic (tool-using) tasks?
- In our benchmarks, qwen3.8-9b-distill-mlx@4bit scores 83.5/100 for agentic workflows. It runs at about 41.7 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 qwen3.8-9b-distill-mlx@4bit for local inference?
- Across 2 community benchmark runs, qwen3.8-9b-distill-mlx@4bit reaches up to 42.2 tok/s and averages 41.7 tok/s, with the fastest results on Apple M1 Max.
- How much memory does qwen3.8-9b-distill-mlx@4bit need?
- The leanest observed configuration used about 7.4 GB of memory (quantizations tested: 4bit).
- Which tools have been used to run qwen3.8-9b-distill-mlx@4bit?
- Benchmarks were submitted using LM Studio. Results are community-contributed and updated as new runs arrive.