Model benchmarks

Qwen3.8-27B-IQ5R-V4-DFlash2-TP2-bf16KV local LLM performance

As of November 2026, Qwen3.8-27B-IQ5R-V4-DFlash2-TP2-bf16KV runs at up to 135.5 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).

radiance
ShareRedditX

Model size

27B

Peak speed

135.5 tok/s

Average speed

129.2 tok/s

Avg PP

3151.1 tok/s

Min memory

n/a

Max context

65,536 tokens

Avg output / run

78,861 tokens

Avg runtime / run

9m 45s

Avg quality

86.6

Benchmark runs

2

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 2 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)
Overall85.786.687.6
Agent Workflow89.391.593.7
Code Generation68.174.580.9
Role Play & Narrative90.893.295.5
Research & Analysis85.387.489.4

Performance by hardware and tool

Every hardware/tool/quantization combination Qwen3.8-27B-IQ5R-V4-DFlash2-TP2-bf16KV has been benchmarked on, ranked by peak token generation speed. Last updated November 2026.

HardwareToolQuantBest tok/sAvg tok/sMemoryContextQualityRuns
AMD Radeon AI PRO R9700radiance—135.5 tok/s129.2 tok/sn/a65,536 tokens86.62

Benchmark runs

All 2 Qwen3.8-27B-IQ5R-V4-DFlash2-TP2-bf16KV 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-bf16KV good for coding?
In our benchmarks, Qwen3.8-27B-IQ5R-V4-DFlash2-TP2-bf16KV scores 74.5/100 for coding. It runs at about 129.2 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-bf16KV good for agentic (tool-using) tasks?
In our benchmarks, Qwen3.8-27B-IQ5R-V4-DFlash2-TP2-bf16KV scores 91.5/100 for agentic workflows. It runs at about 129.2 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-bf16KV for local inference?
Across 2 community benchmark runs, Qwen3.8-27B-IQ5R-V4-DFlash2-TP2-bf16KV reaches up to 135.5 tok/s and averages 129.2 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-bf16KV?
Benchmarks were submitted using radiance. Results are community-contributed and updated as new runs arrive.