Is Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1 good for coding?

In our benchmarks, Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1 scores 65.0/100 for coding. It runs at about 53.8 tok/s, so if you want more speed, ornith-ai/Ornith-1.5-35B-A3B-GGUF is faster (~90.8 tok/s) and still scores well for coding (83.8/100).

Is Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1 good for agentic (tool-using) tasks?

In our benchmarks, Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1 scores 83.4/100 for agentic workflows. It runs at about 53.8 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-oQ4e-mtp-BF16PLE-v1 for local inference?

Across 21 community benchmark runs, Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1 reaches up to 57.9 tok/s and averages 53.8 tok/s, with the fastest results on Apple M4 Max.

How much memory does Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1 need?

The leanest observed configuration used about 78.7 GB of memory.

Which tools have been used to run Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1?

Benchmarks were submitted using oMLX. Results are community-contributed and updated as new runs arrive.

Model benchmarks

Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1 local LLM performance

As of September 2026, Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1 runs at up to 57.9 tok/s for local inference (best of 21 community benchmark runs across 1 GPU).

oMLX

ShareReddit X

Model size

Unknown

Peak speed

57.9 tok/s

Average speed

53.8 tok/s

Avg PP

385.8 tok/s

Min memory

78.7 GB

Max context

131,072 tokens

Avg output / run

59,300 tokens

Avg runtime / run

18m 26s

Avg quality

81.9

Benchmark runs

GPUs tested

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 21 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	74.5	81.9	86.4
Agent Workflow	68.3	83.4	92.9
Code Generation	53.2	65.0	74.6
Role Play & Narrative	84.8	90.3	93.6
Research & Analysis	85.9	88.9	91.3

Performance by hardware and tool

Every hardware/tool/quantization combination Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1 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	—	57.9 tok/s	53.8 tok/s	78.7 GB	131,072 tokens	81.9	21

Benchmark runs

All 21 Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1 runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Open run page →

Frequently asked questions

Is Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1 good for coding?: In our benchmarks, Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1 scores 65.0/100 for coding. It runs at about 53.8 tok/s, so if you want more speed, ornith-ai/Ornith-1.5-35B-A3B-GGUF is faster (~90.8 tok/s) and still scores well for coding (83.8/100).
Is Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1 good for agentic (tool-using) tasks?: In our benchmarks, Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1 scores 83.4/100 for agentic workflows. It runs at about 53.8 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-oQ4e-mtp-BF16PLE-v1 for local inference?: Across 21 community benchmark runs, Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1 reaches up to 57.9 tok/s and averages 53.8 tok/s, with the fastest results on Apple M4 Max.
How much memory does Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1 need?: The leanest observed configuration used about 78.7 GB of memory.
Which tools have been used to run Qwen3.8-Flash-Next-oQ4e-mtp-BF16PLE-v1?: Benchmarks were submitted using oMLX. Results are community-contributed and updated as new runs arrive.