Cursor
AI-native code editor that runs autonomous agents to build, test, and review code
Agentic coding platform with a context engine that indexes large codebases, plus expert agents for review, testing, and PR authoring
Augment Code is an agentic coding platform built around a context engine that indexes large, enterprise-scale codebases. In late 2026 it replaced message-based plans with credit-based pricing: there is no permanent free tier (trial only), Indie is $20/month for 40,000 credits, Standard is $60/month per developer for 130,000 credits per seat, and Max is $200/month for 450,000 credits. Best for engineers working in big, complex repositories who want model-agnostic agents.
Augment Code is an agentic coding platform, branded Cosmos, built around a context engine that indexes large and enterprise-scale codebases so its agents can retrieve the right files and symbols before they edit. Instead of a single autocomplete model, it ships a set of prebuilt expert agents — PR Author, Code Review, Tester, and PR Risk Analysis — that operate across the development lifecycle inside IDEs, from a CLI, and through integrations with GitHub, Slack, Jira, and CI systems. It is model-agnostic: a router called Prism sends each task to a chosen frontier or open-weight model, and teams can bring their own keys.
In late 2026, Augment moved from message-based plans to credit-based pricing. There is no permanent free tier — only a trial with community support — and paid plans start at Indie for $20/month with 40,000 credits, then Standard at $60/month per developer for 130,000 credits per seat, and Max at $200/month per developer for 450,000 credits. Credits reset monthly without rollover, and heavy users top up at $15 per 24,000 credits. A flat $100/month Business plan pools usage across up to 50 seats with SOC 2 Type II, while Enterprise adds private deployments, SSO/SCIM, and an SLA.
Augment Code targets engineers and teams whose main constraint is codebase scale: large monorepos and complex enterprise repositories where the hard part is finding the relevant context, not just generating tokens. The credit model and pooled-seat plans make it most economical for teams rather than occasional solo users.
Starting price: $20/mo · Free tier: no · Model: usage-based
Price history tracked from June 2026
| Plan | Price | Includes |
|---|---|---|
| Trial | Free | Time-limited trial access, no permanent free tier · Community support only · No SLA |
| Indie | $20/mo | 40,000 credits per month for a single user · Auto top-up at $15 per 24,000 extra credits · Credits reset monthly with no rollover · Entry tier for solo developers |
| Standard | $60/dev/mo | 130,000 credits per seat · Up to 20 users with pooled team credits · No per-user caps on the shared pool · Context engine across large repositories |
| Max | $200/dev/mo | 450,000 credits per seat · Up to 20 users with pooled credits · Larger compute headroom for heavy agentic work |
| Business | $100/mo | $100 of pooled usage across models, context engine, and compute · Up to 50 seats with no per-seat charge · SOC 2 Type II, 50 concurrent sessions · CLI, MCP, and native tool support |
| Enterprise | Custom | Unlimited users and sessions · Multi-region compute and private deployments · SSO/OIDC/SCIM, CMEK, ISO 42001, SIEM · Dedicated support and SLA |
| Pros | Cons |
|---|---|
| Context engine is built for genuinely large, enterprise-scale repositories where flat-file context tools struggle | No permanent free tier — only a trial with community support and no SLA, so the lowest ongoing entry is $20/month |
| Model-agnostic routing avoids lock-in to a single LLM provider | Credits do not roll over; heavy use forces paid top-ups at $15 per 24,000 credits, so real cost can exceed the headline price |
| Pooled team credits and SOC 2 Type II / ISO 42001 controls suit regulated organizations | Standalone code completions were sunset, narrowing the product toward agentic workflows rather than lightweight autocomplete |
| Prebuilt expert agents cover review, testing, and PR authoring out of the box | Per-developer pricing on Standard and Max climbs quickly for larger teams |
AI-native code editor that runs autonomous agents to build, test, and review code
AI coding assistant with inline completions, chat, and autonomous agent mode across major IDEs
Agentic AI IDE with Cascade that writes, edits and runs code autonomously inside your editor
Anthropic's AI assistant built for long-context analysis, coding, and safe reasoning
Open-source autonomous coding agent for VS Code that edits files, runs commands, and brings your own model
No. Augment Code offers a time-limited trial but no permanent free tier. The lowest ongoing plan is Indie at $20/month, which includes 40,000 credits for a single user. The trial comes with community support only and no SLA.
Pricing is credit-based as of late 2026, replacing the older message-based model. Indie is $20/month for 40,000 credits, Standard is $60/month per developer for 130,000 credits per seat, and Max is $200/month per developer for 450,000 credits. Credits reset monthly with no rollover, and you can auto top-up at $15 per 24,000 extra credits.
It is a retrieval system that indexes large codebases so the agent can pull the most relevant files and symbols into context before making edits. It is the main reason Augment targets enterprise-scale repositories where simpler context windows fall short.
Augment is model-agnostic. Its Prism router sends each task to a chosen frontier or open-weight model, and you can bring your own keys. This avoids locking your workflow to a single provider.
Augment is an agentic harness that runs frontier models — including Claude — as its engines, so it is not a replacement for Claude but a way to drive it across a large codebase with retrieval and prebuilt experts. For direct code generation and reasoning, Claude remains the stronger choice.
Yes. Alongside managed cloud, Augment offers customer-cloud, VPC, and on-premises deployment, and the Enterprise tier adds SSO/OIDC/SCIM, CMEK, ISO 42001, and SIEM integration for regulated organizations.