Pocket Tank
A 14M-parameter LLM keeps a virtual fish tank alive on an $8 ESP32.
By mediacutlet
- Est. cost
- Cost unknown
- Posted
- GitHub stars
- 207 (GitHub)
Overview
A 14-million-parameter language model makes every decision for a virtual fish tank on an ESP32-S3, with no network: when each fish eats, hides, explores, socialises, rests or flees.
How it works
The model was distilled from a 26B teacher into a 7.56 MB file read straight from flash. Each fish describes its state (hunger, fear, neighbours, personality) as one line of text, and the model completes it with a goal and an urgency. A reflex layer animates that at 25–30 fps on a 1.8-inch AMOLED. You can flash it from the browser to three Waveshare boards.
Hardware
- ESP32-S3
This page is a short summary written by HardwareLab. Build steps, code and photos belong to the creator; follow “View original” for the full project.
More in Local LLM Rigs & Edge Agents
See all 17Esp32 Iphone Agent
A $7 ESP32 lets an AI drive a real iPhone via AirPlay and USB HID.
- ESP32 (USB HID)
- iPhone
IntermediateCost unknown1.4KNeedle 2 Esp32
45M-parameter LLM offline on an ESP32-S3 turning English into valid commands.
- ESP32-S3
AdvancedCost unknown193ESP32s3 LLM Cluster
Seven ESP32-S3s daisy-chained to run a 0.4B ternary LLM.
- 7x ESP32-S3 over SPI
AdvancedCost unknown234