Augment Code pricing, plans and limits

Standard is $20/month flat for up to 50 seats, with a $20/month usage pool shared across LLM calls, Context Engine indexing and compute. Business is $100/month flat, also for up to 50 seats, with a $100/month usage pool. Enterprise is custom-priced for unlimited seats with multi-region compute, SSO/OIDC/SCIM and dedicated support. Every plan allows pay-as-you-go top-ups, which expire 12 months after purchase. There is no permanent free tier; see GitHub Copilot or Windsurf for cheaper single-seat entry points.

The core of the product is the Context Engine, which builds a live semantic index of a codebase instead of relying on keyword search: it maps function-call relationships, API contracts, architectural patterns and cross-service dependencies, indexing up to 500,000 files across 30 or more repositories. That index feeds Cosmos, Augment's agent orchestration layer, and is also exposed over the Model Context Protocol so third-party agents such as Cursor, Claude Code and Codex can query the same context instead of only Augment's own tools. On Scale AI's SWE-bench Pro benchmark, Augment's Auggie agent scored 51.80% in a test the vendor published in February 2026, ahead of Cursor (50.21%), Claude Code (49.75%) and Codex (46.47%) when three of the four agents used the same Claude Opus 4.5 model from Anthropic, a gap the vendor attributes to the Context Engine rather than the underlying model.

Augment Code fits engineering teams working inside large, multi-repository codebases, especially organizations running microservices or maintaining legacy systems where one change can touch dozens of services. Its Remote Agents execute in isolated Ubuntu 22.04 sandboxes, up to ten running in parallel, each with its own repository copy reachable over SSH. Teams already using Claude Code or Cursor (see what Cursor is for if that one's unfamiliar) inside a single repository tend to reach for Augment Code's Context Engine once a codebase outgrows what those tools hold in context, and some connect the two over MCP rather than replacing an existing agent outright. Augment also sells a standalone code review product that reads a pull request against the whole repository instead of just the diff; the vendor's own December 2025 benchmark across five open-source codebases, Sentry, Grafana, Cal.com, Discourse and Keycloak, put it at 65% precision and a 59% F-score.

Augment Code moved to flat, credit-based pricing in October 2025. See the pricing FAQ below for the exact Standard, Business and Enterprise figures; there is no permanent free tier, only limited trial and beta access that the vendor's own pricing page does not size or time-bound. It sits alongside other AI coding assistants on hokai; see the buyer's guide to picking one for how the category breaks down by job rather than brand. Augment runs wherever VS Code or a JetBrains IDE runs, on Windows, macOS or Linux, plus a web dashboard for monitoring Remote Agents; there is no dedicated mobile app.

Leadership changed in 2026: Matt McClernan, previously Augment's chief revenue officer and an alumnus of MongoDB's revenue team, took over as CEO from co-founder-era leader Scott Dietzen in the weeks before May 2026. Compared with IDE-replacement products like Windsurf or GitHub Copilot, Augment has positioned itself away from individual developers: its flat per-team pricing charges a five-person team the same monthly rate as a single engineer, which the vendor frames as a deliberate bet on enterprise engineering organizations rather than the indie-developer market.