Model benchmarks
gemma-4-26B-A4B-it-UD-IQ4_NL local LLM performance
As of September 2026, gemma-4-26B-A4B-it-UD-IQ4_NL runs at up to 93.8 tok/s for local inference (best of 3 community benchmark runs across 1 GPU).
Model size
26B
Peak speed
93.8 tok/s
Average speed
83.2 tok/s
Avg prefill
515.0 tok/s
Min memory
12.7 GB
Max context
131,072 tokens
Avg output / run
18,512 tokens
Avg runtime / run
3m 34s
Avg quality
74.5
Benchmark runs
3
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 3 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 | 73.7 | 74.5 | 75.5 |
| Agent Workflow | 77.4 | 82.5 | 88.3 |
| Code Generation | 64.9 | 67.6 | 71.9 |
| Role Play & Narrative | 74.6 | 80.5 | 84.8 |
| Research & Analysis | 63.1 | 67.5 | 73.2 |
Performance by hardware and tool
Every hardware/tool/quantization combination gemma-4-26B-A4B-it-UD-IQ4_NL has been benchmarked on, ranked by peak token generation speed. Last updated September 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| NVIDIA GeForce RTX 5060 | llama.cpp | — | 93.8 tok/s | 83.2 tok/s | 12.7 GB | 131,072 tokens | 74.5 | 3 |
Frequently asked questions
- Is gemma-4-26B-A4B-it-UD-IQ4_NL good for coding?
- In our benchmarks, gemma-4-26B-A4B-it-UD-IQ4_NL scores 67.6/100 for coding. It runs at about 83.2 tok/s, so if you want more speed, unsloth/Qwen3.8-27B-GGUF:IQ3_S is faster (~104.3 tok/s) and still scores well for coding (83.0/100).
- Is gemma-4-26B-A4B-it-UD-IQ4_NL good for agentic (tool-using) tasks?
- In our benchmarks, gemma-4-26B-A4B-it-UD-IQ4_NL scores 82.5/100 for agentic workflows. It runs at about 83.2 tok/s, so if you want more speed, Tiel-Coder-35B-A3B-Q4_K_S is faster (~165.0 tok/s) and still scores well for agentic workflows (89.3/100).
- How fast is gemma-4-26B-A4B-it-UD-IQ4_NL for local inference?
- Across 3 community benchmark runs, gemma-4-26B-A4B-it-UD-IQ4_NL reaches up to 93.8 tok/s and averages 83.2 tok/s, with the fastest results on NVIDIA GeForce RTX 5060.
- How much memory does gemma-4-26B-A4B-it-UD-IQ4_NL need?
- The leanest observed configuration used about 12.7 GB of memory.
- Which tools have been used to run gemma-4-26B-A4B-it-UD-IQ4_NL?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.