Model benchmarks
smollm3-3b local LLM performance
As of October 2026, smollm3-3b runs at up to 74.9 tok/s for local inference (best of 3 community benchmark runs across 2 GPUs).
Model size
3B
Peak speed
74.9 tok/s
Average speed
65.1 tok/s
Avg PP
2117.2 tok/s
Min memory
3.0 GB
Max context
32,768 tokens
Avg output / run
18,611 tokens
Avg runtime / run
5m 1s
Avg quality
34.8
Benchmark runs
3
GPUs tested
2
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 | 27.4 | 34.8 | 41.8 |
| Agent Workflow | 30.2 | 48.8 | 67.0 |
| Code Generation | 0.6 | 4.7 | 7.6 |
| Role Play & Narrative | 48.4 | 54.5 | 58.4 |
| Research & Analysis | 26.6 | 31.3 | 36.7 |
Performance by hardware and tool
Every hardware/tool/quantization combination smollm3-3b 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 Laptop GPU | Bionic | — | 74.9 tok/s | 74.6 tok/s | 5.7 GB | 32,768 tokens | 39.0 | 2 |
| NVIDIA GeForce RTX 2050 | LM Studio | Q4_K_M | 46.2 tok/s | 46.2 tok/s | 3.0 GB | 8,192 tokens | 26.5 | 1 |
Benchmark runs
All 3 smollm3-3b runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is smollm3-3b good for coding?
- In our benchmarks, smollm3-3b scores 4.7/100 for coding. It runs at about 65.1 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 smollm3-3b good for agentic (tool-using) tasks?
- In our benchmarks, smollm3-3b scores 48.8/100 for agentic workflows. It runs at about 65.1 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 smollm3-3b for local inference?
- Across 3 community benchmark runs, smollm3-3b reaches up to 74.9 tok/s and averages 65.1 tok/s, with the fastest results on NVIDIA GeForce RTX 3060 Laptop GPU.
- How much memory does smollm3-3b need?
- The leanest observed configuration used about 3.0 GB of memory (quantizations tested: Q4_K_M).
- Which tools have been used to run smollm3-3b?
- Benchmarks were submitted using Bionic, LM Studio. Results are community-contributed and updated as new runs arrive.