Model benchmarks

Qwen3.8-27B-IQ5R-V4-DFlash2-TP2 local LLM performance

As of November 2026, Qwen3.8-27B-IQ5R-V4-DFlash2-TP2 runs at up to 137.9 tok/s for local inference (best of 11 community benchmark runs across 1 GPU).

radiance
ShareRedditX

Model size

27B

Peak speed

137.9 tok/s

Average speed

128.8 tok/s

Avg PP

3309.6 tok/s

Min memory

n/a

Max context

65,536 tokens

Avg output / run

78,081 tokens

Avg runtime / run

9m 48s

Avg quality

88.1

Benchmark runs

11

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 11 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)
Overall85.288.191.1
Agent Workflow86.590.492.6
Code Generation72.380.088.7
Role Play & Narrative91.494.296.3
Research & Analysis85.087.790.0

Performance by hardware and tool

Every hardware/tool/quantization combination Qwen3.8-27B-IQ5R-V4-DFlash2-TP2 has been benchmarked on, ranked by peak token generation speed. Last updated November 2026.

HardwareToolQuantBest tok/sAvg tok/sMemoryContextQualityRuns
AMD Radeon AI PRO R9700radiance—137.9 tok/s128.8 tok/sn/a65,536 tokens88.111

Benchmark runs

All 11 Qwen3.8-27B-IQ5R-V4-DFlash2-TP2 runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Frequently asked questions

Is Qwen3.8-27B-IQ5R-V4-DFlash2-TP2 good for coding?
In our benchmarks, Qwen3.8-27B-IQ5R-V4-DFlash2-TP2 scores 80.0/100 for coding. It runs at about 128.8 tok/s, so if you want more speed, ornith-1.5-35b-apex is faster (~149.2 tok/s) and still scores well for coding (83.5/100).
Is Qwen3.8-27B-IQ5R-V4-DFlash2-TP2 good for agentic (tool-using) tasks?
In our benchmarks, Qwen3.8-27B-IQ5R-V4-DFlash2-TP2 scores 90.4/100 for agentic workflows. It runs at about 128.8 tok/s, so if you want more speed, Ornith-1.5-35B-A3B-APEX-MTP-I-Quality-MTP-ngram is faster (~150.8 tok/s) and still scores well for agentic workflows (90.1/100).
How fast is Qwen3.8-27B-IQ5R-V4-DFlash2-TP2 for local inference?
Across 11 community benchmark runs, Qwen3.8-27B-IQ5R-V4-DFlash2-TP2 reaches up to 137.9 tok/s and averages 128.8 tok/s, with the fastest results on AMD Radeon AI PRO R9700.
Which tools have been used to run Qwen3.8-27B-IQ5R-V4-DFlash2-TP2?
Benchmarks were submitted using radiance. Results are community-contributed and updated as new runs arrive.