Model benchmarks
nemotron-bolt:THINKING local LLM performance
As of October 2026, nemotron-bolt:THINKING runs at up to 135.8 tok/s for local inference (best of 2 community benchmark runs across 0 GPUs).
Model size
Unknown
Peak speed
135.8 tok/s
Average speed
133.0 tok/s
Avg PP
1602.7 tok/s
Min memory
n/a
Max context
131,072 tokens
Avg output / run
7,817 tokens
Avg runtime / run
1m 5s
Avg quality
67.6
Benchmark runs
2
GPUs tested
0
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 | 66.1 | 67.6 | 69.0 |
| Agent Workflow | 75.3 | 75.7 | 76.1 |
| Code Generation | 63.1 | 64.0 | 64.9 |
| Role Play & Narrative | 72.9 | 73.3 | 73.7 |
| Research & Analysis | 50.7 | 57.3 | 63.8 |
Performance by hardware and tool
Every hardware/tool/quantization combination nemotron-bolt:THINKING 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 |
|---|---|---|---|---|---|---|---|---|
| AMD Radeon Graphics | llama.cpp | — | 135.8 tok/s | 133.0 tok/s | n/a | 131,072 tokens | 67.6 | 2 |
Benchmark runs
All 2 nemotron-bolt:THINKING runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is nemotron-bolt:THINKING good for coding?
- In our benchmarks, nemotron-bolt:THINKING scores 64.0/100 for coding. It runs at about 133.0 tok/s, so if you want more speed, Qwen3.6-35B-A3B-Q4_K_XL is faster (~156.9 tok/s) and still scores well for coding (82.2/100).
- Is nemotron-bolt:THINKING good for agentic (tool-using) tasks?
- In our benchmarks, nemotron-bolt:THINKING scores 75.7/100 for agentic workflows. It runs at about 133.0 tok/s, so if you want more speed, swift-1.5-qwen3.8-flash-next-iq3_s is faster (~259.0 tok/s) and still scores well for agentic workflows (88.9/100).
- How fast is nemotron-bolt:THINKING for local inference?
- Across 2 community benchmark runs, nemotron-bolt:THINKING reaches up to 135.8 tok/s and averages 133.0 tok/s, with the fastest results on AMD Radeon Graphics.
- Which tools have been used to run nemotron-bolt:THINKING?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.