The Stack v3 releases 5T code tokens across 700+ programming languages
LoubnaBenAllal1 · x · 2026-07-23
The Stack team says it is releasing The Stack v3 at a time when strong open models are needed more than ever.
- The dataset contains 5T tokens of code.
- It spans 700+ programming languages.
- The release is positioned as infrastructure for building and training stronger open code models.
Related event: Hugging Face Releases The Stack v3: 114TB Largest Open Source Code Dataset(13 posts)→
More from Research
- Token Collapse: Study Finds Graph Soft Prompting Tokens in LLMs Fail to Encode Structure — Sapienza · 2026-10-27
- UniAction: unified action conditioning lets one world model handle mixed action types — ZeYanjie · 2026-10-27
- Infini-AI Lab open-sources Hermit, a framework for agent harness evolution and continual learning — BeidiChen · 2026-10-27
- Meta FAIR's Self-Direction turns agent failures into reusable LoRA skills, +13.8pp on Terminal-Bench — karen_ullrich · 2026-10-27
- DONIC hits 86.79% on bimanual dexterous tracking, crushing prior SOTA's 9.43% — liuziwei7 · 2026-10-27
- Why don't more LLMs ship 10M-token context? A researcher asks about attention tradeoffs — agarwl_ · 2026-10-27