Model benchmarks
google/gemma-4-26b-a4b-qat local LLM performance
As of July 2026, google/gemma-4-26b-a4b-qat runs at up to 51.4 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
LM Studio
Model size
26B
Peak speed
51.4 tok/s
Average speed
33.5 tok/s
Min memory
12.7 GB
Max context
9,384 tokens
Best quality
87.0
Benchmark runs
2
GPUs tested
1
Performance by hardware and tool
Every hardware/tool/quantization combination google/gemma-4-26b-a4b-qat has been benchmarked on, ranked by peak token generation speed. Last updated July 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| NVIDIA GeForce RTX 4070 Ti SUPER | LM Studio | — | 51.4 tok/s | 51.4 tok/s | 14.6 GB | 9,384 tokens | 73.7 | 1 |
| CPU only | LM Studio | — | 15.7 tok/s | 15.7 tok/s | 12.7 GB | 6,372 tokens | 76.6 | 1 |
Frequently asked questions
- How fast is google/gemma-4-26b-a4b-qat for local inference?
- Across 2 community benchmark runs, google/gemma-4-26b-a4b-qat reaches up to 51.4 tok/s and averages 33.5 tok/s, with the fastest results on NVIDIA GeForce RTX 4070 Ti SUPER.
- How much memory does google/gemma-4-26b-a4b-qat need?
- The leanest observed configuration used about 12.7 GB of memory.
- Which tools have been used to run google/gemma-4-26b-a4b-qat?
- Benchmarks were submitted using LM Studio. Results are community-contributed and updated as new runs arrive.