Augment Code Review 2026: Pooled Usage Pricing, Pros and Cons

Augment Code changed what it sells this year. The old credit tiers are gone from its pricing page, replaced by two flat team plans that sit on top of a metered usage pool, and the headline product is now Cosmos, a cloud platform for agents that take a ticket to a reviewed pull request. Toolradar data: Toolradar, the software directory we run, tracks 179 AI coding tools as of October 19, 2026, and the median cheapest paid plan among the 58 that publish a price is $19 a month, per its AI coding ranking. Augment's Standard plan sits next to that median, except it covers a team rather than one seat.

This review explains how the pool bills, prices a 10-developer team at three spending levels, and sets the result against Claude Team, GitHub Copilot and Devin. Augment figures come from its pricing page and its token-based pricing docs, read on October 19, 2026. For the wider field, start with our AI coding assistants ranking.

Augment Code is an agentic coding platform built around its Context Engine, which indexes a codebase so agents can work across large repositories. Today it ships as Cosmos, a web platform where prebuilt Experts such as a PR author, a code review fleet and a work dispatcher run tickets through implementation, review and verification, plus the Auggie CLI for terminal and automation work. Augment's docs still list clients for VS Code, JetBrains IDEs and Vim or Neovim, a Context Engine MCP server and GitHub pull request review.

The buyer is an engineering lead at a team of 5 to 50 who wants agents to take whole tickets, not only autocomplete lines, and who is comfortable paying per token. A solo developer who wants a flat-price assistant in one editor is better served by a seat plan; our Claude Code review and Windsurf review cover two of them.

An admin picks Standard or Business, invites up to 50 teammates, and everyone draws from one balance. Augment's FAQ counts a seat as any member with workspace access, and neither plan charges per seat, so an occasional user costs nothing until they run a task. Above 50 people, the only route is Enterprise.

Every request costs three things. Model tokens bill at the provider's public API list price, a flat service fee is added to that LLM spend to fund the Context Engine and Cosmos, and Cosmos sandbox time bills by the hour in 5-minute increments, rounded up. When the included usage runs out, billing continues pay-as-you-go at the same rates with no minimum top-up, and purchased top-ups stay valid for 12 months. A dashboard shows spend per member, per model and per component, so a lead can see who is draining the pool.

Augment Code sells two flat team plans, Standard and Business, each covering up to 50 seats and including usage equal to its fee, with Enterprise priced by quote. Usage is model tokens at list price plus the service fee and compute, and top-ups bill at the same rates. Prices were read on October 19, 2026 on the Augment pricing page and the token-based pricing docs.

What the pool buys: with no compute, Business's included usage covers about $71.43 of raw model tokens, because the service fee takes the rest (our division). Augment's docs price a Sonnet-class bug fix at about $0.25, so the pool covers roughly 400 such tasks a month for the whole team, or 40 per developer on a team of 10. On Standard, the same team gets about 8 such tasks per developer before top-ups.

Scenario, a 10-developer team, by our arithmetic. At $50 of model tokens a month, usage is $70: Standard bills $70 and Business stays at its flat fee. At $400 of tokens, usage is $560, and both plans bill $560, since the fee is only a floor. At $1,000 of tokens, both bill $1,400. For comparison, ten Claude Team standard seats cost $200 a month on annual billing, ten Devin team seats $480, and ten Copilot Pro licences $100. Those are seat plans with usage allowances rather than metered bills, so they cap the bill and Augment does not.

Methodology: list prices in USD before tax, read on the Augment, Claude, Devin and GitHub Copilot pricing pages on October 19, 2026, multiplied by seat count, with the service fee applied as Augment's docs describe. We did not run these tools; token spend levels are illustrative inputs, not measurements.

Buy Augment for a team of 5 to 50 where a few people run agents hard and the rest dip in. The pool and the no-seat model reward uneven usage, the Context Engine pays off on large repositories, and Cosmos suits teams that want tickets turned into reviewed pull requests rather than suggestions in an editor. Our AI coding agents ranking compares it with other agent-first tools.

Skip it if your finance team needs a fixed monthly number. Metered tokens plus the service fee mean the bill tracks usage with no ceiling except the budget you watch, while seat plans such as Claude Team or Copilot cap spend per person. Our AI coding assistant pricing comparison lines up the seat models, and the enterprise AI coding ranking covers the procurement side.

Wait and watch if you are signing a long contract. Harness acquired select Augment assets on October 8, 2026, and says Cosmos will become the Harness Cosmos Software Factory Agent, so plan names, pricing and integrations may shift. JetBrains users should also read our JetBrains AI assistants ranking, since the pricing page names Cosmos and the CLI but not the IDE clients.