Model benchmarks
Qwen3.8-Flash-Next-Q5GU-Q6Attn-g64-BF16-PLE-MLX-MTP-Q8 local LLM performance
As of September 2026, Qwen3.8-Flash-Next-Q5GU-Q6Attn-g64-BF16-PLE-MLX-MTP-Q8 runs at up to 52.3 tok/s for local inference (best of 3 community benchmark runs across 1 GPU).
Model size
4B
Peak speed
52.3 tok/s
Average speed
51.8 tok/s
Avg prefill
347.5 tok/s
Min memory
88.1 GB
Max context
262,144 tokens
Avg output / run
13,047 tokens
Avg runtime / run
4m 41s
Avg quality
82.9
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 | 77.3 | 82.9 | 87.4 |
| Agent Workflow | 73.9 | 84.1 | 92.1 |
| Code Generation | 69.1 | 76.6 | 81.5 |
| Role Play & Narrative | 81.7 | 86.2 | 91.8 |
| Research & Analysis | 84.3 | 84.7 | 85.2 |
Performance by hardware and tool
Every hardware/tool/quantization combination Qwen3.8-Flash-Next-Q5GU-Q6Attn-g64-BF16-PLE-MLX-MTP-Q8 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 |
|---|---|---|---|---|---|---|---|---|
| Apple M4 Max | oMLX | — | 52.3 tok/s | 51.8 tok/s | 88.1 GB | 262,144 tokens | 82.9 | 3 |
Frequently asked questions
- Is Qwen3.8-Flash-Next-Q5GU-Q6Attn-g64-BF16-PLE-MLX-MTP-Q8 good for coding?
- In our benchmarks, Qwen3.8-Flash-Next-Q5GU-Q6Attn-g64-BF16-PLE-MLX-MTP-Q8 scores 76.6/100 for coding. It runs at about 51.8 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-Flash-Next-Q5GU-Q6Attn-g64-BF16-PLE-MLX-MTP-Q8 good for agentic (tool-using) tasks?
- In our benchmarks, Qwen3.8-Flash-Next-Q5GU-Q6Attn-g64-BF16-PLE-MLX-MTP-Q8 scores 84.1/100 for agentic workflows. It runs at about 51.8 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-Flash-Next-Q5GU-Q6Attn-g64-BF16-PLE-MLX-MTP-Q8 for local inference?
- Across 3 community benchmark runs, Qwen3.8-Flash-Next-Q5GU-Q6Attn-g64-BF16-PLE-MLX-MTP-Q8 reaches up to 52.3 tok/s and averages 51.8 tok/s, with the fastest results on Apple M4 Max.
- How much memory does Qwen3.8-Flash-Next-Q5GU-Q6Attn-g64-BF16-PLE-MLX-MTP-Q8 need?
- The leanest observed configuration used about 88.1 GB of memory.
- Which tools have been used to run Qwen3.8-Flash-Next-Q5GU-Q6Attn-g64-BF16-PLE-MLX-MTP-Q8?
- Benchmarks were submitted using oMLX. Results are community-contributed and updated as new runs arrive.