Jetson AGX Orin / Llama 3.2 3B
Can a Jetson AGX Orin run Llama 3.2 3B?
Yes, at a comfortable chat speed. Measured generation speed: 80.4 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 |
|---|---|---|---|---|---|---|
| Jetson AGX Orin Jetson AGX Orin | INT4 | MLC 0.1.2 (dustynv/mlc:0.1.2-r36.3.0) | 80.4 | – | – | source ↗ accessed 2026-10-17 |
Llama 3.2 3B on other boards
- Jetson Orin NX 8GB Super 46.5 tok/s best
- Jetson Orin NX 16GB Super 45.8 tok/s best
- Jetson Orin Nano Super 43.1 tok/s best
- Jetson Orin NX 16GB 40.4 tok/s best
- Jetson Orin NX 8GB 34.5 tok/s best
- Jetson Orin Nano 27.7 tok/s best
Buy a Jetson AGX Orin
- Jetson AGX Orin Developer Kit: $3499 raised from $1,999, source ↗ (2026-07-21) Check price on AmazonBuy at Seeed Studio
Ran Llama 3.2 3B on a Jetson AGX Orin? Submit your numbers →