Is gemma4:12b-it-qat good for coding?

In our benchmarks, gemma4:12b-it-qat scores 62.6/100 for coding. It runs at about 37.0 tok/s, so if you want more speed, nvidia/Qwen3.8-27B-NVFP4 is faster (~63.3 tok/s) and still scores well for coding (82.7/100).

Is gemma4:12b-it-qat good for agentic (tool-using) tasks?

In our benchmarks, gemma4:12b-it-qat scores 76.9/100 for agentic workflows. It runs at about 37.0 tok/s, so if you want more speed, swift-iq2_xs is faster (~91.5 tok/s) and still scores well for agentic workflows (91.4/100).

How fast is gemma4:12b-it-qat for local inference?

Across 6 community benchmark runs, gemma4:12b-it-qat reaches up to 54.5 tok/s and averages 37.0 tok/s, with the fastest results on AMD Radeon RX 7800 XT.

How much memory does gemma4:12b-it-qat need?

The leanest observed configuration used about 6.5 GB of memory (quantizations tested: Q4_0).

Which tools have been used to run gemma4:12b-it-qat?

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

Model benchmarks

gemma4:12b-it-qat local LLM performance

As of October 2026, gemma4:12b-it-qat runs at up to 54.5 tok/s for local inference (best of 6 community benchmark runs across 3 GPUs).

OllamaQ4_0

ShareReddit X

Model size

11.9B

Peak speed

54.5 tok/s

Average speed

37.0 tok/s

Avg PP

824.2 tok/s

Min memory

6.5 GB

Max context

65,536 tokens

Avg output / run

11,255 tokens

Avg runtime / run

6m 47s

Avg quality

71.4

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	68.0	71.4	77.1
Agent Workflow	67.5	76.9	84.1
Code Generation	52.2	62.6	66.4
Role Play & Narrative	70.9	75.8	82.8
Research & Analysis	63.0	70.1	76.9

Performance by hardware and tool

Every hardware/tool/quantization combination gemma4:12b-it-qat 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 7800 XT	Ollama	Q4_0	54.5 tok/s	52.6 tok/s	7.4 GB	65,536 tokens	68.7	3
AMD Radeon RX 9050 / 9060 XT	Ollama	Q4_0	33.5 tok/s	33.5 tok/s	8.2 GB	65,536 tokens	74.6	1
NVIDIA GeForce GTX 1080	Ollama	Q4_0	15.3 tok/s	15.2 tok/s	6.5 GB	8,192 tokens	73.7	2

Benchmark runs

All 6 gemma4:12b-it-qat runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.

Open run page →

Frequently asked questions

Is gemma4:12b-it-qat good for coding?: In our benchmarks, gemma4:12b-it-qat scores 62.6/100 for coding. It runs at about 37.0 tok/s, so if you want more speed, nvidia/Qwen3.8-27B-NVFP4 is faster (~63.3 tok/s) and still scores well for coding (82.7/100).
Is gemma4:12b-it-qat good for agentic (tool-using) tasks?: In our benchmarks, gemma4:12b-it-qat scores 76.9/100 for agentic workflows. It runs at about 37.0 tok/s, so if you want more speed, swift-iq2_xs is faster (~91.5 tok/s) and still scores well for agentic workflows (91.4/100).
How fast is gemma4:12b-it-qat for local inference?: Across 6 community benchmark runs, gemma4:12b-it-qat reaches up to 54.5 tok/s and averages 37.0 tok/s, with the fastest results on AMD Radeon RX 7800 XT.
How much memory does gemma4:12b-it-qat need?: The leanest observed configuration used about 6.5 GB of memory (quantizations tested: Q4_0).
Which tools have been used to run gemma4:12b-it-qat?: Benchmarks were submitted using Ollama. Results are community-contributed and updated as new runs arrive.