Model benchmarks
Gemma4-31B-IT_t1p95k64m05 local LLM performance
As of November 2026, Gemma4-31B-IT_t1p95k64m05 runs at up to 51.4 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
31B
Peak speed
51.4 tok/s
Average speed
51.4 tok/s
Avg PP
556.8 tok/s
Min memory
54.5 GB
Max context
32,768 tokens
Avg output / run
9,220 tokens
Avg runtime / run
2m 58s
Avg quality
75.1
Benchmark runs
2
GPUs tested
1
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 | 73.5 | 75.1 | 76.7 |
| Agent Workflow | 80.2 | 82.3 | 84.4 |
| Code Generation | 69.7 | 72.2 | 74.7 |
| Role Play & Narrative | 68.0 | 71.3 | 74.7 |
| Research & Analysis | 73.0 | 74.5 | 76.1 |
Performance by hardware and tool
Every hardware/tool/quantization combination Gemma4-31B-IT_t1p95k64m05 has been benchmarked on, ranked by peak token generation speed. Last updated November 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| Tesla V100-PCIE-32GB | llama.cpp | — | 51.4 tok/s | 51.4 tok/s | 54.5 GB | 32,768 tokens | 75.1 | 2 |
Benchmark runs
All 2 Gemma4-31B-IT_t1p95k64m05 runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is Gemma4-31B-IT_t1p95k64m05 good for coding?
- In our benchmarks, Gemma4-31B-IT_t1p95k64m05 scores 72.2/100 for coding. It runs at about 51.4 tok/s, so if you want more speed, local-qwen is faster (~60.2 tok/s) and still scores well for coding (84.8/100).
- Is Gemma4-31B-IT_t1p95k64m05 good for agentic (tool-using) tasks?
- In our benchmarks, Gemma4-31B-IT_t1p95k64m05 scores 82.3/100 for agentic workflows. It runs at about 51.4 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 Gemma4-31B-IT_t1p95k64m05 for local inference?
- Across 2 community benchmark runs, Gemma4-31B-IT_t1p95k64m05 reaches up to 51.4 tok/s and averages 51.4 tok/s, with the fastest results on Tesla V100-PCIE-32GB.
- How much memory does Gemma4-31B-IT_t1p95k64m05 need?
- The leanest observed configuration used about 54.5 GB of memory.
- Which tools have been used to run Gemma4-31B-IT_t1p95k64m05?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.