Model benchmarks
Gemma4:E4B/QAT-MTP@131K local LLM performance
As of August 2026, Gemma4:E4B/QAT-MTP@131K runs at up to 347.5 tok/s for local inference (best of 2 community benchmark runs across 0 GPUs).
Model size
4B
Peak speed
347.5 tok/s
Average speed
303.9 tok/s
Min memory
n/a
Max context
16,384 tokens
Avg runtime / run
33s
Avg quality
67.8
Benchmark runs
2
GPUs tested
0
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 2 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 | 65.7 | 67.8 | 70.0 |
| Agent Workflow | 70.7 | 71.7 | 72.7 |
| Code Generation | 61.1 | 66.9 | 72.7 |
| Role Play & Narrative | 65.8 | 67.6 | 69.3 |
| Research & Analysis | 63.2 | 65.1 | 67.0 |
Performance by hardware and tool
Every hardware/tool/quantization combination Gemma4:E4B/QAT-MTP@131K has been benchmarked on, ranked by peak token generation speed. Last updated August 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| CPU only | LM Studio | — | 347.5 tok/s | 303.9 tok/s | n/a | 16,384 tokens | 67.8 | 2 |
Benchmark runs
All 2 Gemma4:E4B/QAT-MTP@131K runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- How fast is Gemma4:E4B/QAT-MTP@131K for local inference?
- Across 2 community benchmark runs, Gemma4:E4B/QAT-MTP@131K reaches up to 347.5 tok/s and averages 303.9 tok/s.
- Which tools have been used to run Gemma4:E4B/QAT-MTP@131K?
- Benchmarks were submitted using LM Studio. Results are community-contributed and updated as new runs arrive.