GitHub Codespaces

GitHub Codespaces puts a full VS Code environment in the cloud, provisioned from a devcontainer.json that lives in the repo. Teams get identical environments for every developer, one-click launches from pull requests, prebuilds that cut startup to seconds, and JetBrains Gateway support for IntelliJ and PyCharm users. It integrates natively with GitHub Actions, secrets, and organization policies for managing fleets of environments at scale.

Pricing includes a free tier of 60 core-hours per month on a 2-core machine with 15 GB of storage, then usage-based billing at roughly $0.18 per core-hour plus storage. The main limitations are cost management at team scale and ML fit: GPU machine types that existed in limited beta appear discontinued in current docs, so heavy model training still belongs on dedicated GPU infrastructure. 2026-era reviews rate it highly for web and backend development but mediocre for daily deep-learning work.

For AI engineering teams, Codespaces pairs naturally with Copilot: the same cloud environment that runs the code can host the agent loop, and gh codespace CLI commands make environment management scriptable from CI and automation.

From any GitHub repo click Code > Codespaces, or use gh codespace create from the GitHub CLI.