Model benchmarks
NVIDIA-Nemotron-3-Super-120B-A12B-OptiQ-2bit local LLM performance
As of July 2026, NVIDIA-Nemotron-3-Super-120B-A12B-OptiQ-2bit runs at up to 41.6 tok/s for local inference (best of 3 community benchmark runs across 1 GPU).
Model size
120B
Peak speed
41.6 tok/s
Average speed
37.1 tok/s
Min memory
n/a
Max context
180,000 tokens
Avg quality
64.4
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 | 50.3 | 64.4 | 73.0 |
| Agent Workflow | 57.8 | 68.6 | 76.0 |
| Code Generation | 60.1 | 62.0 | 63.5 |
| Role Play & Narrative | 8.0 | 54.0 | 81.5 |
| Research & Analysis | 69.8 | 73.0 | 75.7 |
Performance by hardware and tool
Every hardware/tool/quantization combination NVIDIA-Nemotron-3-Super-120B-A12B-OptiQ-2bit has been benchmarked on, ranked by peak token generation speed. Last updated July 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| Apple M5 Max | oMLX | — | 41.6 tok/s | 37.1 tok/s | n/a | 180,000 tokens | 64.4 | 3 |
Benchmark runs
All 3 NVIDIA-Nemotron-3-Super-120B-A12B-OptiQ-2bit runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is NVIDIA-Nemotron-3-Super-120B-A12B-OptiQ-2bit good for coding?
- In our benchmarks, NVIDIA-Nemotron-3-Super-120B-A12B-OptiQ-2bit scores 62.0/100 for coding. It runs at about 37.1 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 NVIDIA-Nemotron-3-Super-120B-A12B-OptiQ-2bit good for agentic (tool-using) tasks?
- In our benchmarks, NVIDIA-Nemotron-3-Super-120B-A12B-OptiQ-2bit scores 68.6/100 for agentic workflows. It runs at about 37.1 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 NVIDIA-Nemotron-3-Super-120B-A12B-OptiQ-2bit for local inference?
- Across 3 community benchmark runs, NVIDIA-Nemotron-3-Super-120B-A12B-OptiQ-2bit reaches up to 41.6 tok/s and averages 37.1 tok/s, with the fastest results on Apple M5 Max.
- Which tools have been used to run NVIDIA-Nemotron-3-Super-120B-A12B-OptiQ-2bit?
- Benchmarks were submitted using oMLX. Results are community-contributed and updated as new runs arrive.