Paper
An Evaluation of LLMs Inference on Popular Single-board Computers
External resource. Not made or verified by Public Protocol.
A 2025 paper testing 25 quantised open models on Raspberry Pi 4, Raspberry Pi 5 and Orange Pi 5 Pro with Ollama and Llamafile. It reports 5 to 15 tokens a second for 1 to 1.5 billion parameter models and 2 to 5 for 3 billion on the Pi 5, peak draws of 8 to 14 W, and up to four times the throughput with Llamafile.
https://arxiv.org/abs/2511.07425
- Maintainer
- Tung Nguyen and Tuyen Nguyen (arXiv)
- Licence, as stated at its source
- arXiv non-exclusive distribution licence 1.0
- Open or commercial
- Open
- Maturity
- PrototypeOur reading, 16 November 2026
- Safety
- S0
- Review
- Listed 16 November 2026by Public Protocol editors
- Checked
- 16 November 2026
- Next check due
- 16 May 2027
- Steward
- Public Protocol editors
Where it sits in the blueprint
- Domains
- Stages
Design
- Elements it supports
- Context
Climate, setting, scale, cost and skill: not stated in this entry. Check the source against your own place and numbers.
Verification
No verification is recorded on Public Platform for this entry. Records are added when someone reads it for accuracy, reviews it with a stated qualification, builds it or measures it.
Record
- How it came in
Seed catalogue, 16 November 2026. Listed in the edge computing research of 2026-11-16. The source was opened and described on the date shown.
- Licence
arXiv non-exclusive distribution licence 1.0. Read at https://arxiv.org/abs/2511.07425. How licences are recorded
- Something wrong?