Model benchmarks
model local LLM performance
As of October 2026, model runs at up to 84.1 tok/s for local inference (best of 2 community benchmark runs across 1 GPU).
Model size
1.7B
Peak speed
84.1 tok/s
Average speed
68.9 tok/s
Avg PP
1326.1 tok/s
Min memory
5.1 GB
Max context
65,536 tokens
Avg output / run
44,127 tokens
Avg runtime / run
10m 46s
Avg quality
72.7
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 | 62.6 | 72.7 | 82.8 |
| Agent Workflow | 66.8 | 73.4 | 79.9 |
| Code Generation | 43.9 | 62.1 | 80.2 |
| Role Play & Narrative | 58.9 | 70.9 | 83.0 |
| Research & Analysis | 80.8 | 84.4 | 88.0 |
Performance by hardware and tool
Every hardware/tool/quantization combination model 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 GeForce RTX 3060 | llama.cpp | — | 84.1 tok/s | 68.9 tok/s | 5.1 GB | 65,536 tokens | 72.7 | 2 |
Benchmark runs
All 2 model runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is model good for coding?
- In our benchmarks, model scores 62.1/100 for coding. It runs at about 68.9 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 model good for agentic (tool-using) tasks?
- In our benchmarks, model scores 73.4/100 for agentic workflows. It runs at about 68.9 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 model for local inference?
- Across 2 community benchmark runs, model reaches up to 84.1 tok/s and averages 68.9 tok/s, with the fastest results on NVIDIA GeForce RTX 3060.
- How much memory does model need?
- The leanest observed configuration used about 5.1 GB of memory.
- Which tools have been used to run model?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.