Is qwen3.5:4b good for coding?

In our benchmarks, qwen3.5:4b scores 33.0/100 for coding. It runs at about 62.2 tok/s, so if you want more speed, Tiel-Coder-35B-A3B.APEX-I-MiniPlus-V2.1 is faster (~71.6 tok/s) and still scores well for coding (81.8/100).

Is qwen3.5:4b good for agentic (tool-using) tasks?

In our benchmarks, qwen3.5:4b scores 43.2/100 for agentic workflows. It runs at about 62.2 tok/s, so if you want more speed, Nex-N2.5-mini-APEX-Mini is faster (~88.2 tok/s) and still scores well for agentic workflows (91.3/100).

How fast is qwen3.5:4b for local inference?

Across 6 community benchmark runs, qwen3.5:4b reaches up to 102.3 tok/s and averages 62.2 tok/s, with the fastest results on NVIDIA GeForce RTX 3060 Ti.

How much memory does qwen3.5:4b need?

The leanest observed configuration used about 3.1 GB of memory (quantizations tested: Q4_K_M).

Which tools have been used to run qwen3.5:4b?

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

Model benchmarks

qwen3.5:4b local LLM performance

As of October 2026, qwen3.5:4b runs at up to 102.3 tok/s for local inference (best of 6 community benchmark runs across 4 GPUs).

BionicOllamaQ4_K_M

ShareReddit X

Model size

4.7B

Peak speed

102.3 tok/s

Average speed

62.2 tok/s

Avg PP

1615.7 tok/s

Min memory

3.1 GB

Max context

65,536 tokens

Avg output / run

18,323 tokens

Avg runtime / run

11m 41s

Avg quality

51.8

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 6 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	40.1	51.8	61.4
Agent Workflow	8.4	43.2	63.4
Code Generation	22.8	33.0	47.5
Role Play & Narrative	71.0	77.2	82.9
Research & Analysis	24.9	53.7	68.6

Performance by hardware and tool

Every hardware/tool/quantization combination qwen3.5:4b 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
NVIDIA GeForce RTX 3060 Ti	Ollama	Q4_K_M	102.3 tok/s	102.3 tok/s	3.1 GB	8,192 tokens	36.8	1
AMD Radeon RX 7800 XT	Ollama	Q4_K_M	96.4 tok/s	60.5 tok/s	5.2 GB	65,536 tokens	54.9	2
NVIDIA GeForce RTX 3060 Laptop GPU	Bionic	—	70.6 tok/s	70.2 tok/s	6.7 GB	16,384 tokens	57.1	2
NVIDIA GeForce GTX 1050 Ti	Ollama	Q4_K_M	9.4 tok/s	9.4 tok/s	3.6 GB	8,192 tokens	49.9	1

Benchmark runs

All 6 qwen3.5:4b runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Open run page →

Frequently asked questions

Is qwen3.5:4b good for coding?: In our benchmarks, qwen3.5:4b scores 33.0/100 for coding. It runs at about 62.2 tok/s, so if you want more speed, Tiel-Coder-35B-A3B.APEX-I-MiniPlus-V2.1 is faster (~71.6 tok/s) and still scores well for coding (81.8/100).
Is qwen3.5:4b good for agentic (tool-using) tasks?: In our benchmarks, qwen3.5:4b scores 43.2/100 for agentic workflows. It runs at about 62.2 tok/s, so if you want more speed, Nex-N2.5-mini-APEX-Mini is faster (~88.2 tok/s) and still scores well for agentic workflows (91.3/100).
How fast is qwen3.5:4b for local inference?: Across 6 community benchmark runs, qwen3.5:4b reaches up to 102.3 tok/s and averages 62.2 tok/s, with the fastest results on NVIDIA GeForce RTX 3060 Ti.
How much memory does qwen3.5:4b need?: The leanest observed configuration used about 3.1 GB of memory (quantizations tested: Q4_K_M).
Which tools have been used to run qwen3.5:4b?: Benchmarks were submitted using Bionic, Ollama. Results are community-contributed and updated as new runs arrive.