karpathy/nanochat
The best ChatGPT that $100 can buy.
Builder
karpathy
karpathy • individual
Stars
54,381
Using upstream star count
Forks
7,359
Using upstream fork count
Open Issues
0
Activity Score
0/100
0 commits in 30d
Created
Oct 13, 2025
Project creation date
nanochat is the simplest experimental harness for training LLMs. It is designed to run on a single GPU node, the code is minimal/hackable, and it covers all major LLM stages including tokenization, pretraining, finetuning, evaluation, inference, and a chat UI. For example, you can train your own GPT-2 capability LLM (which cost ~$43,000 to train in 2019) for only $48 (~2 hours of 8XH100 GPU node) and then talk to it in a familiar ChatGPT-like web UI. On a spot instance, the total cost can be clo
Unmapped
category
Deployment Context
Modalities
Skill Areas
tag
Updated 2 months ago
7 Days
0
30 Days
0
90 Days
8
submit new time to GPT-2 leaderboard entry: 99 minutes
Andrej Karpathy • Mar 14, 2026
Autoresearch round 2: smear, backout, and hyperparameter tuning
Andrej Karpathy • Mar 14, 2026
new leaderboard entry coming from improvements of autoresearch round 1, time to gpt-2 from 2.02 hour
Andrej Karpathy • Mar 10, 2026
pgvector cosine similarity · $0
Loading…