Specs at a glance
| CPU | Broadcom BCM2712, 4× Cortex-A76 @ 2.4 GHz |
| RAM options | 1 / 2 / 4 / 8 / 16 GB LPDDR4X |
| Storage | microSD or NVMe via M.2 HAT |
| Power draw | 5-12 W |
| Form factor | 85 × 56 mm credit-card SBC |
| Local LLM capability | Small models only (≤ 3B Q4) |
| Price point | €80-110 |
Choosing between devices? The hardware buyer's guide compares them by budget, and Local LLM benchmarks has the published speeds, machine by machine.
Overview
The Raspberry Pi 5 is the de-facto first device for pocket AI in 2026. The 8 GB SKU has enough RAM for Hermes Agent without browser automation, or for Nanobot with everything. NVMe via the official M.2 HAT removes the SD card bottleneck. Raspberry Pi gives a typical bare-board current of 800 mA, about 4 W, before storage and load. Local LLM inference is the limit: the Slips project measured about 11 tokens per second with 1B models and about 5 with 3B models on a Pi 5 (Ollama, Q4_K_M), so small models are the ceiling, and they trail Claude meaningfully on agentic tasks. For agents that delegate the LLM call to a cloud API, the Pi 5 is competitive with €30/month VPS plans.
Best for
- First-time self-hosters wanting low power + always-on
- Hermes Agent without browser tool
- Nanobot deployments
- MCP server hosts
Not for
- Local LLM inference beyond ~3B param Q4 quants
- Browser automation (Chromium too heavy)
- Multi-user agent workloads
Compatible self-hosted agents
Listed as compatible with Raspberry Pi 5 from each agent's requirements, not from a test (with the caveats from “Best for” / “Not for” above):
Where to buy
Manufacturer page: https://www.raspberrypi.com/products/raspberry-pi-5/. The Amazon link is an affiliate link: it costs you nothing and funds the site. Details on our disclosure page.
See also: all AI hardware · AI agents tracker · AI CVEs · edge AI hardware buyer's guide · scan your AI stack · our methodology.