Famed gamer creates working 5 million parameter ChatGPT AI model in Minecraft, made with 439 million blocks — AI trained to hold conversations, working model runs inference in the game
CraftGPT is hewn from 1020x260x1656 Redstone blocks in the game.

A developer and enthusiast Minecrafter has showcased a project dubbed CraftGPT on GitHub. In an amazing feat of Minecraft Redstone engineering, Sammyuri — famed for building a 1Hz CPU inside the game — has built a small language model that runs on a computer inside Minecraft, trained on the TinyChat dataset. The CraftGPT project is hewn from 1,020 x 260 x 1,656 blocks in the game (439 million in total), and functions as advertised, but a major usability drawback is that you will have to “wait a couple [of] hours for the response to be generated.”
The video above shows the colossal in-game computer that was built to run CraftGPT. Sammyuri explains that the game’s Distant Horizons mod was needed to capture this ‘fly-by’ footage of the computer build.
Naturally, even with such an impressive Redstone project, CraftGPT isn’t going to replace more traditional methods of getting an LLM up and running on your PC. Sammyuri asks users/testers to temper their expectations. “The model is very prone to going off-topic, producing responses that are not grammatically correct, or simply outputting garbage,” admits the project developer.
So, how did Sammyuri use Redstone to put this project together? Redstone provides electronic components within the Minecraft environment. The video shows the in-game CraftGPT contraption being put together component-by-component. It has tokenizers, matrix multipliers, and so on. Sammyuri explains that the small language model used was created without command blocks or data packs in Minecraft. Moreover, “the model has 5,087,280 parameters, trained in Python on the TinyChat dataset of basic English conversations,” says the developer.
AI-wranglers may also be interested to know that the 5-million parameter CraftGPT “has an embedding dimension of 240, vocabulary of 1920 tokens, and consists of 6 layers. The context window size is 64 tokens, which is enough for (very) short conversations.” Sammyuri adds that “Most weights were quantized to 8 bits, although the embedding and LayerNorm weights are stored at 18 and 24 bits respectively.”
The need for speed
CraftGPT is admittedly not in any way a rival to the chatbot sites and apps you may use day to day. We already mentioned some limitations of chat quality, depth, and even accuracy. But probably by far its largest wrinkle is performance-related.
A Minecraft Redstone constructed processor or computer might be a modern marvel, but it pales in comparison to real hardware. Thus, the CraftGPT chat experience is severely hampered by response time. Typically, the system will need about two hours to answer a simple prompt, says Sammyuri. This is even with the tick rate increased to about 40,000x speed using MCHPRS (the Minecraft High Performance Redstone Server).
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
Previously mind-blowing Minecraft Redstone feats include standalone 16-bit CPUs and the IRIS Computer, which was capable of running a version of DOOM (1993) in Minecraft.
Follow Tom's Hardware on Google News, or add us as a preferred source, to get our up-to-date news, analysis, and reviews in your feeds. Make sure to click the Follow button!

Mark Tyson is a news editor at Tom's Hardware. He enjoys covering the full breadth of PC tech; from business and semiconductor design to products approaching the edge of reason.