Model benchmarks
ddalcu/Qwen3.8-Flash-Next-MLX-Serve-mixed-4-8bit local LLM performance
As of September 2026, ddalcu/Qwen3.8-Flash-Next-MLX-Serve-mixed-4-8bit runs at up to 72.0 tok/s for local inference (best of 3 community benchmark runs across 1 GPU).
Model size
Unknown
Peak speed
72.0 tok/s
Average speed
71.5 tok/s
Avg PP
1910.9 tok/s
Min memory
n/a
Max context
262.144 tokens
Avg output / run
15.625 tokens
Avg runtime / run
3m 23s
Avg quality
70.6
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 | 68.0 | 70.6 | 74.8 |
| Agent Workflow | 69.9 | 72.1 | 74.3 |
| Code Generation | 38.4 | 52.5 | 66.9 |
| Role Play & Narrative | 73.4 | 79.0 | 86.3 |
| Research & Analysis | 74.5 | 78.9 | 83.4 |
Performance by hardware and tool
Every hardware/tool/quantization combination ddalcu/Qwen3.8-Flash-Next-MLX-Serve-mixed-4-8bit 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 | vLLM | — | 72.0 tok/s | 71.5 tok/s | n/a | 262.144 tokens | 70.6 | 3 |
Benchmark runs
All 3 ddalcu/Qwen3.8-Flash-Next-MLX-Serve-mixed-4-8bit runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is ddalcu/Qwen3.8-Flash-Next-MLX-Serve-mixed-4-8bit good for coding?
- In our benchmarks, ddalcu/Qwen3.8-Flash-Next-MLX-Serve-mixed-4-8bit scores 52.5/100 for coding. It runs at about 71.5 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 ddalcu/Qwen3.8-Flash-Next-MLX-Serve-mixed-4-8bit good for agentic (tool-using) tasks?
- In our benchmarks, ddalcu/Qwen3.8-Flash-Next-MLX-Serve-mixed-4-8bit scores 72.1/100 for agentic workflows. It runs at about 71.5 tok/s, so if you want more speed, Ornith-1.5-35B-A3B-BigBang-oQ8e-mtp is faster (~102.7 tok/s) and still scores well for agentic workflows (91.1/100).
- How fast is ddalcu/Qwen3.8-Flash-Next-MLX-Serve-mixed-4-8bit for local inference?
- Across 3 community benchmark runs, ddalcu/Qwen3.8-Flash-Next-MLX-Serve-mixed-4-8bit reaches up to 72.0 tok/s and averages 71.5 tok/s, with the fastest results on Apple M4 Max.
- How much memory does ddalcu/Qwen3.8-Flash-Next-MLX-Serve-mixed-4-8bit need?
- The leanest observed configuration used about n/a of memory.
- Which tools have been used to run ddalcu/Qwen3.8-Flash-Next-MLX-Serve-mixed-4-8bit?
- Benchmarks were submitted using vLLM. Results are community-contributed and updated as new runs arrive.