Model benchmarks
LFM3-1.7B-BF16 local LLM performance
As of November 2026, LFM3-1.7B-BF16 runs at up to 83.5 tok/s for local inference (best of 3 community benchmark runs across 2 GPUs).
Model size
1.7B
Peak speed
83.5 tok/s
Average speed
50.4 tok/s
Avg PP
1450.5 tok/s
Min memory
4.1 GB
Max context
262,144 tokens
Avg output / run
43,534 tokens
Avg runtime / run
16m 6s
Avg quality
62.0
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 | 57.4 | 62.0 | 67.7 |
| Agent Workflow | 70.9 | 73.0 | 76.3 |
| Code Generation | 25.2 | 35.6 | 50.7 |
| Role Play & Narrative | 58.1 | 62.2 | 66.3 |
| Research & Analysis | 71.8 | 77.3 | 82.4 |
Performance by hardware and tool
Every hardware/tool/quantization combination LFM3-1.7B-BF16 has been benchmarked on, ranked by peak token generation speed. Last updated November 2026.
| Hardware | Tool | Quant | Best tok/s | Avg tok/s | Memory | Context | Quality | Runs |
|---|---|---|---|---|---|---|---|---|
| NVIDIA GeForce RTX 3060 | llama.cpp | — | 83.5 tok/s | 83.5 tok/s | 7.6 GB | 262,144 tokens | 68.5 | 1 |
| NVIDIA GeForce RTX 4060 Laptop GPU | llama.cpp | — | 35.5 tok/s | 33.9 tok/s | 4.1 GB | 32,768 tokens | 58.7 | 2 |
Benchmark runs
All 3 LFM3-1.7B-BF16 runs submitted so far — expand one for its hardware, quality breakdown and per-scenario detail.
Frequently asked questions
- Is LFM3-1.7B-BF16 good for coding?
- In our benchmarks, LFM3-1.7B-BF16 scores 35.6/100 for coding. It runs at about 50.4 tok/s, so if you want more speed, nvidia/Qwen3.8-27B-NVFP4 is faster (~63.3 tok/s) and still scores well for coding (82.7/100).
- Is LFM3-1.7B-BF16 good for agentic (tool-using) tasks?
- In our benchmarks, LFM3-1.7B-BF16 scores 73.0/100 for agentic workflows. It runs at about 50.4 tok/s, so if you want more speed, swift-iq2_xs is faster (~91.5 tok/s) and still scores well for agentic workflows (91.4/100).
- How fast is LFM3-1.7B-BF16 for local inference?
- Across 3 community benchmark runs, LFM3-1.7B-BF16 reaches up to 83.5 tok/s and averages 50.4 tok/s, with the fastest results on NVIDIA GeForce RTX 3060.
- How much memory does LFM3-1.7B-BF16 need?
- The leanest observed configuration used about 4.1 GB of memory.
- Which tools have been used to run LFM3-1.7B-BF16?
- Benchmarks were submitted using llama.cpp. Results are community-contributed and updated as new runs arrive.