Model benchmarks
gemma4-26b-opus-distill local LLM performance
As of October 2026, gemma4-26b-opus-distill runs at up to 63.5 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
26B
Peak speed
63.5 tok/s
Average speed
63.3 tok/s
Avg PP
824.1 tok/s
Min memory
22.6 GB
Max context
65,536 tokens
Avg output / run
5,308 tokens
Avg runtime / run
1m 42s
Avg quality
68.0
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 | 65.7 | 68.0 | 70.2 |
| Agent Workflow | 64.1 | 71.8 | 79.4 |
| Code Generation | 55.8 | 56.9 | 57.9 |
| Role Play & Narrative | 74.7 | 77.0 | 79.2 |
| Research & Analysis | 64.4 | 66.4 | 68.3 |
Performance by hardware and tool
Every hardware/tool/quantization combination gemma4-26b-opus-distill 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 |
|---|---|---|---|---|---|---|---|---|
| NVIDIA RTX A2000 12GB | llama.cpp | — | 63.5 tok/s | 63.3 tok/s | 22.6 GB | 65,536 tokens | 68.0 | 2 |
Benchmark runs
All 2 gemma4-26b-opus-distill runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is gemma4-26b-opus-distill good for coding?
- In our benchmarks, gemma4-26b-opus-distill scores 56.9/100 for coding. It runs at about 63.3 tok/s, so if you want more speed, Nex-N2.5-mini-IQ3_XXS is faster (~91.4 tok/s) and still scores well for coding (81.6/100).
- Is gemma4-26b-opus-distill good for agentic (tool-using) tasks?
- In our benchmarks, gemma4-26b-opus-distill scores 71.8/100 for agentic workflows. It runs at about 63.3 tok/s, so if you want more speed, Nex-N2.5-mini-APEX-Mini is faster (~88.2 tok/s) and still scores well for agentic workflows (91.3/100).
- How fast is gemma4-26b-opus-distill for local inference?
- Across 2 community benchmark runs, gemma4-26b-opus-distill reaches up to 63.5 tok/s and averages 63.3 tok/s, with the fastest results on NVIDIA RTX A2000 12GB.
- How much memory does gemma4-26b-opus-distill need?
- The leanest observed configuration used about 22.6 GB of memory.
- Which tools have been used to run gemma4-26b-opus-distill?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.