Amp is Sourcegraph’s AI coding agent, rebuilt around a fast CLI and deep code intelligence that understands how your codebase actually fits together.
Amp combines frontier models with Sourcegraph’s heritage in code search and intelligence. The agent plans and executes changes with awareness of code structure across repositories, and its Neo CLI rebuild focused on speed, quality of edits and a frictionless loop from prompt to passing tests.
Amp supports autonomous modes for longer tasks, human-in-the-loop review for sensitive changes, and parallel agent threads for working multiple problems at once. It integrates with editors and CI, and its pricing approach aims to make high-quality agent usage predictable for individuals and teams. Plan-first execution produces reviewable diffs tied back to intent, which makes agent output easier to trust in large repositories. Free-tier access invites teams to validate quality against their own codebase before committing.
Amp offers a free tier for trying the agent and light usage, with usage-based paid plans that scale with model consumption. Team and enterprise plans add shared configuration, security features and support. Usage-based pricing aligns cost with model consumption, and predictable billing matters for teams running agents continuously.
Amp is for developers and teams working in large codebases where structural understanding matters, and for engineers who want strong agent capability without a heavyweight IDE. It also suits developers working across multiple repositories where structural awareness of the whole estate is essential.
Yes, Amp includes a free tier that lets you evaluate real work before paying. Ongoing heavy use moves to usage-based pricing aligned with model consumption.
Amp can be driven from the CLI and integrated into automation pipelines, and Sourcegraph offers programmatic access for enterprise workflows and internal tooling.
Amp’s edge is code intelligence on large repositories, where its structural awareness outperforms agents that only see files. It is younger than market leaders but competitive on quality and increasingly popular with serious engineering teams. Amp earns its place on big, interconnected codebases where shallower agents lose the thread; smaller projects may not see the difference.
Gemini CLI is Google's open-source AI coding agent that brings frontier Gemini models into the terminal with a remarkably generous free tier.OverviewLaunched…
OpenCode is an open-source, terminal-native coding agent that is model-agnostic, letting you run any LLM provider or local model against your repository.OverviewOpenCode…
Crush is Charm's glamorous open-source coding agent for the terminal, bringing multi-model AI assistance to a beautifully crafted command-line experience.OverviewCrush comes from…