EdgeRaspberry Pi

Raspberry Pi 5 8GB

Curated Aggregate·2025-06-222 workloads · 3 records
MyAI RatingNot scoredInsufficient comparison evidence

VRAM

0 GB

TDP

12 W

MSRP

$80

Perf/W

0.35 tok/s/W

Cost/1K tok

$0.20/M

Tested

2025-06-22

Quick answer

How fast is Raspberry Pi 5 8GB for local AI workloads?

Raspberry Pi 5 8GB has a source-attributed result of 1.8 tok/s on Mistral 7B (batch 1, 2048-token context, Q4_K_M; runtime not documented). This is a reference report, not an independently verified lab result. A 70B Q4_K_M artifact needs roughly 40GB for weights alone; smaller memory configurations require explicit offload or model splitting and do not establish full-GPU residency.

Source: MyAIHardware benchmark database (bench-rpi5-mistral)As of 2025-07-04

Overview

Embedded Baseline

Raspberry Pi 5 (8GB) — ARM-based single-board computer with quad-core Cortex-A76 at 2.4GHz, 8GB LPDDR4X SDRAM, VideoCore VII GPU. 12W peak power via USB-C. The standard platform for embedded computing and IoT.

AI Usefulness

Runs 3B Q4 models at ~3-5 tok/s CPU-only. Sufficient for tiny chatbots, keyword classification, and basic embedding tasks. Not suitable for 7B+ models at usable speeds. Best for learning, edge prototyping, and ultra-low-power AI deployments where a $80 budget and 12W power draw are the constraints.

Verdict

Raspberry Pi 5 8GB with 0GB VRAM at 12W TDP, scored across 2 workloads with 3 benchmark records.

Reference workload

Mistral 7B

1.8 tok/s

Quantization

Q4_K_M

2K context · batch 1

LLM Inference Performance

02468Phi-3 MiniMistral 7B

Benchmarks (2 workloads)

WorkloadScoreQuantContextσStatusTested
Phi-3 Mini

llm

4.2tok/sQ4_K_M2K, Curated Aggregate2025-06-22
Mistral 7B

llm

1.8tok/sQ4_K_M2K, Curated Aggregate2025-07-04

MyAI Score: not scored

There is insufficient comparable evidence to score Raspberry Pi 5 8GB. A score needs results for multiple devices with matching workload, runtime version, quantization, context and batch size.

Workload Fit

Planning estimates for 0GB: weights plus at least 2GB or 10% overhead. Actual KV cache depends on model, context and cache format; confirm the artifact before buying.

Q4
Q8
FP16
7-8B
Won't fit
Won't fit
Won't fit
13-14B
Won't fit
Won't fit
Won't fit
32B
Won't fit
Won't fit
Won't fit
70B
Won't fit
Won't fit
Won't fit
Reported batch-one examples
Mistral 7B2 tok/s
Phi-3 Mini4 tok/s

Source

Slow but functional — proves local LLM on a $80 board.

View sourceHow we benchmark →

Public Trust Layer

Trust score

6/10

MyAI rating

Not scored

Runs

Not documented

Freshness

Stale

Source-linked row with explicit verification status.

Record date: 2025-07-04; 432 days old.

Open primary source

Best place to start

Where to buy

Search first

Retailer we'd check first

Amazon search

For mainstream AI hardware, Amazon usually updates street pricing, seller availability, and shipping speed faster than most comparison sites.

Search Amazon listings
  • +Use $80 as your price anchor unless the part is clearly supply-constrained or newly launched.
  • +Check the exact cooler, board partner, or memory configuration before buying. The silicon may match, but noise and thermals do not.
  • +If the Amazon price looks inflated, wait or compare against recent street pricing rather than paying a panic premium.

There is insufficient comparable evidence for a purchase recommendation. Check model fit, software support and a current seller quote.

Affiliate note: we do not have a device-level ASIN yet, so this opens tagged Amazon search results for the exact product name.

Same Tier

NVIDIA Jetson AGX Orin 64GB
64GB
NVIDIA Jetson Orin Nano 8GB
8GB
Raspberry Pi 5 + AI HAT+ (Hailo-8)
8GB
NVIDIA Jetson Orin Nano Super 8GB
8GB
NVIDIA Jetson AGX Orin 64GB
64GB
NVIDIA Jetson Orin NX 16GB
16GB