Raspberry Pi 5 + AI HAT+ 2 / Llama 3.2 1B
Can a Raspberry Pi 5 + AI HAT+ 2 run Llama 3.2 1B?
Yes, usable but slower than reading speed. Measured generation speed: 8.7 tok/s across 1 public run.
Rule of thumb: 10+ tok/s feels like a chat; 4–10 is usable for assistants and automations; under 4 is batch-job territory. Speeds depend on quantization, runtime, context length and cooling. Check each source.
| Board | Quant | Runtime | Gen tok/s | Prompt tok/s | TTFT ms | Source |
|---|---|---|---|---|---|---|
| Raspberry Pi 5 + AI HAT+ 2 Raspberry Pi 5 + Hailo-10H | Hailo-compiled | hailo-ollama | 8.7 | – | – | source ↗ published 2026 |
Llama 3.2 1B on other boards
- Jetson AGX Orin 163.9 tok/s best
- Jetson Orin Nano 54.8 tok/s best
- Raspberry Pi 5 11.22 tok/s best
- Jetson Orin Nano Super 10.96 tok/s best
- M5Stack Module LLM 4.48 tok/s best
Buy a Raspberry Pi 5 + AI HAT+ 2
- Raspberry Pi AI HAT+ 2 (add-on, needs a Pi 5): $180 official list price after Apr 2026 increase (launched at $130), source ↗ (2026-04-01) Check price (affiliate slot)
- Raspberry Pi 5 + AI HAT+ 2 (as tested in paper): $451.83 price quoted in arXiv paper, source ↗ (2026-04-24) Check price (affiliate slot)
Ran Llama 3.2 1B on a Raspberry Pi 5 + AI HAT+ 2? Submit your numbers →