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

In our benchmarks, Qwen3.8-Flash-Next-oQ4e-mtp scores 71.7/100 for coding. It runs at about 58.0 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 good for agentic (tool-using) tasks?

In our benchmarks, Qwen3.8-Flash-Next-oQ4e-mtp scores 83.0/100 for agentic workflows. It runs at about 58.0 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 for local inference?

Across 100 community benchmark runs, Qwen3.8-Flash-Next-oQ4e-mtp reaches up to 132.5 tok/s and averages 58.0 tok/s, with the fastest results on Apple M5 Ultra.

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

The leanest observed configuration used about 46.6 GB of memory (quantizations tested: oQ4e).

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

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

Model benchmarks

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

As of October 2026, Qwen3.8-Flash-Next-oQ4e-mtp runs at up to 132.5 tok/s for local inference (best of 100 community benchmark runs across 4 GPUs).

oMLXoQ4e

ShareReddit X

Model size

125B

Peak speed

132.5 tok/s

Average speed

58.0 tok/s

Avg PP

989.3 tok/s

Min memory

46.6 GB

Max context

262,144 tokens

Avg output / run

17,073 tokens

Avg runtime / run

7m 18s

Avg quality

81.8

Benchmark runs

100

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 100 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	76.5	81.8	85.3
Agent Workflow	71.7	83.0	91.3
Code Generation	58.7	71.7	80.8
Role Play & Narrative	75.7	87.9	95.2
Research & Analysis	78.0	84.5	89.4

Performance by hardware and tool

Every hardware/tool/quantization combination Qwen3.8-Flash-Next-oQ4e-mtp has been benchmarked on, ranked by peak token generation speed. Last updated October 2026.

Hardware	Tool	Quant	Best tok/s	Avg tok/s	Memory	Context	Quality	Runs
Apple M5 Ultra	oMLX	oQ4e	132.5 tok/s	132.5 tok/s	102.0 GB	65,536 tokens	79.2	1
Apple M4 Max	oMLX	oQ4e	75.5 tok/s	71.2 tok/s	70.5 GB	262,144 tokens	81.6	16
Apple M4 Max	oMLX	—	67.6 tok/s	58.1 tok/s	70.0 GB	182,272 tokens	81.9	74
Apple M5 Max	oMLX	oQ4e	27.9 tok/s	24.6 tok/s	46.6 GB	150,000 tokens	81.4	4
Apple M2 Ultra	oMLX	—	27.2 tok/s	25.2 tok/s	68.9 GB	131,072 tokens	80.9	5

Benchmark runs

All 100 Qwen3.8-Flash-Next-oQ4e-mtp 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 good for coding?: In our benchmarks, Qwen3.8-Flash-Next-oQ4e-mtp scores 71.7/100 for coding. It runs at about 58.0 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 good for agentic (tool-using) tasks?: In our benchmarks, Qwen3.8-Flash-Next-oQ4e-mtp scores 83.0/100 for agentic workflows. It runs at about 58.0 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 for local inference?: Across 100 community benchmark runs, Qwen3.8-Flash-Next-oQ4e-mtp reaches up to 132.5 tok/s and averages 58.0 tok/s, with the fastest results on Apple M5 Ultra.
How much memory does Qwen3.8-Flash-Next-oQ4e-mtp need?: The leanest observed configuration used about 46.6 GB of memory (quantizations tested: oQ4e).
Which tools have been used to run Qwen3.8-Flash-Next-oQ4e-mtp?: Benchmarks were submitted using oMLX. Results are community-contributed and updated as new runs arrive.