How fast is qwen2.5-0.5b-nnl-distilled-clean-q4_k_m for local inference?

Across 3 community benchmark runs, qwen2.5-0.5b-nnl-distilled-clean-q4_k_m reaches up to 361.0 tok/s and averages 346.3 tok/s, with the fastest results on AMD Radeon RX 7900 XTX.

How much memory does qwen2.5-0.5b-nnl-distilled-clean-q4_k_m need?

The leanest observed configuration used about 1.5 GB of memory (quantizations tested: Q4_K_M).

Which tools have been used to run qwen2.5-0.5b-nnl-distilled-clean-q4_k_m?

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

Model benchmarks

qwen2.5-0.5b-nnl-distilled-clean-q4_k_m local LLM performance

As of October 2026, qwen2.5-0.5b-nnl-distilled-clean-q4_k_m runs at up to 361.0 tok/s for local inference (best of 3 community benchmark runs across 1 GPU).

llama.cppQ4_K_M

ShareReddit X

Model size

0.5B

Peak speed

361.0 tok/s

Average speed

346.3 tok/s

Avg PP

10580.4 tok/s

Min memory

1.5 GB

Max context

8,192 tokens

Avg output / run

11,424 tokens

Avg runtime / run

33s

Avg quality

6.7

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 3 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	4.6	6.7	8.3
Agent Workflow	2.9	10.1	15.2
Code Generation	0.1	3.4	8.5
Role Play & Narrative	4.9	7.7	12.2
Research & Analysis	1.8	5.5	7.9

Performance by hardware and tool

Every hardware/tool/quantization combination qwen2.5-0.5b-nnl-distilled-clean-q4_k_m 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
AMD Radeon RX 7900 XTX	llama.cpp	Q4_K_M	361.0 tok/s	346.3 tok/s	1.5 GB	8,192 tokens	6.7	3

Benchmark runs

All 3 qwen2.5-0.5b-nnl-distilled-clean-q4_k_m runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Open run page →

Frequently asked questions

How fast is qwen2.5-0.5b-nnl-distilled-clean-q4_k_m for local inference?: Across 3 community benchmark runs, qwen2.5-0.5b-nnl-distilled-clean-q4_k_m reaches up to 361.0 tok/s and averages 346.3 tok/s, with the fastest results on AMD Radeon RX 7900 XTX.
How much memory does qwen2.5-0.5b-nnl-distilled-clean-q4_k_m need?: The leanest observed configuration used about 1.5 GB of memory (quantizations tested: Q4_K_M).
Which tools have been used to run qwen2.5-0.5b-nnl-distilled-clean-q4_k_m?: Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.