Is qwen3.6:latest good for coding?

In our benchmarks, qwen3.6:latest scores 63.6/100 for coding. It runs at about 38.2 tok/s, so if you want more speed, unsloth/Qwen3.8-27B-GGUF:IQ3_S is faster (~104.3 tok/s) and still scores well for coding (83.0/100).

Is qwen3.6:latest good for agentic (tool-using) tasks?

In our benchmarks, qwen3.6:latest scores 79.3/100 for agentic workflows. It runs at about 38.2 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 qwen3.6:latest for local inference?

Across 12 community benchmark runs, qwen3.6:latest reaches up to 142.1 tok/s and averages 38.2 tok/s, with the fastest results on NVIDIA GeForce RTX 5080.

How much memory does qwen3.6:latest need?

The leanest observed configuration used about 22.9 GB of memory (quantizations tested: Q4_K_M).

Which tools have been used to run qwen3.6:latest?

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

Model benchmarks

qwen3.6:latest local LLM performance

As of September 2026, qwen3.6:latest runs at up to 142.1 tok/s for local inference (best of 12 community benchmark runs across 2 GPUs).

OllamaQ4_K_M

ShareReddit X

Model size

36.0B

Peak speed

142.1 tok/s

Average speed

38.2 tok/s

Avg PP

690.3 tok/s

Min memory

22.9 GB

Max context

180.000 tokens

Avg output / run

15.297 tokens

Avg runtime / run

Avg quality

76.7

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 12 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	69.3	76.7	82.4
Agent Workflow	61.9	79.3	89.0
Code Generation	51.4	63.6	77.8
Role Play & Narrative	73.7	85.9	91.6
Research & Analysis	70.7	77.9	84.5

Performance by hardware and tool

Every hardware/tool/quantization combination qwen3.6:latest has been benchmarked on, ranked by peak token generation speed. Last updated September 2026.

Hardware	Tool	Quant	Best tok/s	Avg tok/s	Memory	Context	Quality	Runs
NVIDIA GeForce RTX 5080	Ollama	Q4_K_M	142.1 tok/s	142.1 tok/s	22.9 GB	65.536 tokens	73.9	1
AMD Radeon RX 7900 XTX	Ollama	Q4_K_M	83.0 tok/s	28.8 tok/s	22.9 GB	180.000 tokens	76.9	11

Benchmark runs

All 12 qwen3.6:latest runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Frequently asked questions

Is qwen3.6:latest good for coding?: In our benchmarks, qwen3.6:latest scores 63.6/100 for coding. It runs at about 38.2 tok/s, so if you want more speed, unsloth/Qwen3.8-27B-GGUF:IQ3_S is faster (~104.3 tok/s) and still scores well for coding (83.0/100).
Is qwen3.6:latest good for agentic (tool-using) tasks?: In our benchmarks, qwen3.6:latest scores 79.3/100 for agentic workflows. It runs at about 38.2 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 qwen3.6:latest for local inference?: Across 12 community benchmark runs, qwen3.6:latest reaches up to 142.1 tok/s and averages 38.2 tok/s, with the fastest results on NVIDIA GeForce RTX 5080.
How much memory does qwen3.6:latest need?: The leanest observed configuration used about 22.9 GB of memory (quantizations tested: Q4_K_M).
Which tools have been used to run qwen3.6:latest?: Benchmarks were submitted using Ollama. Results are community-contributed and updated as new runs arrive.