Model benchmarks

Qwen3.8-Flash-Next-REAP-384-oQ5e-BF16-MTP-PLE local LLM performance

As of September 2026, Qwen3.8-Flash-Next-REAP-384-oQ5e-BF16-MTP-PLE runs at up to 62.1 tok/s for local inference (best of 22 community benchmark runs across 1 GPU).

oMLXoQ5e
ShareRedditX

Model size

Unknown

Peak speed

62.1 tok/s

Average speed

61.1 tok/s

Avg PP

536.4 tok/s

Min memory

70.0 GB

Max context

262.144 tokens

Avg output / run

11.363 tokens

Avg runtime / run

3m 11s

Avg quality

82.5

Benchmark runs

22

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 22 runs. The low and high columns show how much the judge’s score varies between runs, and need at least two runs to display.

TaskP5 (low)AvgP95 (high)
Overall79.182.585.5
Agent Workflow75.882.990.4
Code Generation66.574.582.6
Role Play & Narrative81.488.894.2
Research & Analysis77.883.888.4

Performance by hardware and tool

Every hardware/tool/quantization combination Qwen3.8-Flash-Next-REAP-384-oQ5e-BF16-MTP-PLE has been benchmarked on, ranked by peak token generation speed. Last updated September 2026.

HardwareToolQuantBest tok/sAvg tok/sMemoryContextQualityRuns
Apple M4 MaxoMLXoQ5e62.1 tok/s61.1 tok/s70.0 GB262.144 tokens82.522

Benchmark runs

All 22 Qwen3.8-Flash-Next-REAP-384-oQ5e-BF16-MTP-PLE runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Frequently asked questions

Is Qwen3.8-Flash-Next-REAP-384-oQ5e-BF16-MTP-PLE good for coding?
In our benchmarks, Qwen3.8-Flash-Next-REAP-384-oQ5e-BF16-MTP-PLE scores 74.5/100 for coding. It runs at about 61.1 tok/s, so if you want more speed, ornith-ai/Ornith-1.5-35B-A3B-GGUF is faster (~96.3 tok/s) and still scores well for coding (83.4/100).
Is Qwen3.8-Flash-Next-REAP-384-oQ5e-BF16-MTP-PLE good for agentic (tool-using) tasks?
In our benchmarks, Qwen3.8-Flash-Next-REAP-384-oQ5e-BF16-MTP-PLE scores 82.9/100 for agentic workflows. It runs at about 61.1 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-REAP-384-oQ5e-BF16-MTP-PLE for local inference?
Across 22 community benchmark runs, Qwen3.8-Flash-Next-REAP-384-oQ5e-BF16-MTP-PLE reaches up to 62.1 tok/s and averages 61.1 tok/s, with the fastest results on Apple M4 Max.
How much memory does Qwen3.8-Flash-Next-REAP-384-oQ5e-BF16-MTP-PLE need?
The leanest observed configuration used about 70.0 GB of memory (quantizations tested: oQ5e).
Which tools have been used to run Qwen3.8-Flash-Next-REAP-384-oQ5e-BF16-MTP-PLE?
Benchmarks were submitted using oMLX. Results are community-contributed and updated as new runs arrive.