Cursor AI In-Depth Review: Navigating the New Subscription Tiers, Bugbot, and Cloud Agents

Cursor AI has evolved from a clever VS Code fork into the command center of modern AI-native software engineering. By indexing entire repositories locally and offering deep Model Context Protocol (MCP) integrations, it allows developers to orchestrate complex multi-file edits effortlessly. However, major shifts in how Cursor bills for third-party frontier models have forced solo developers and engineering managers to rethink their subscription strategies.

Core Features That Keep Developers Hooked

1. Unlimited Tab & Auto Mode: Cursor’s proprietary inline autocomplete (Tab) and its cost-efficient Auto routing mode remain effectively unlimited on paid plans, handling boilerplate code with near-zero latency.
2. Cloud Agents & Composer: Developers can spin up background Cloud Agents to tackle refactoring tasks, generate unit tests, and resolve dependencies off their local machine while they focus on core architecture.
3. Bugbot Automated PR Reviews: Moving away from its legacy $40/month flat seat fee, Cursor’s Bugbot now operates on a usage-based model averaging $1.00 to $1.50 per pull request review, scanning diffs for logic flaws and security regressions before human reviewers even open GitHub.

The Pricing Restructure Explained

Because frontier models burn through tokens at wildly different rates depending on task complexity, Cursor transitioned from counting "fast requests" to a dollar-credit pool and tiered structure:

1. Free Plan ($0/month): Limited trial quota to test inline completions and basic agentic workflows.
2. Plus Plan ($8/month): An ultra-low entry tier with one strict caveat: it is restricted exclusively to Grok models inside the editor, with no access to Claude or GPT.
3. Premium / Pro ($20–$30/month): The standard setup for professional coders, unlocking frontier models from Anthropic, OpenAI, and Google, alongside included Bugbot checks.
4. Pro+ / Super ($60–$100/month) & Ultra ($200/month): Designed for heavy AI-native engineers running autonomous agents continuously throughout the workday, offering up to 20x–50x the base credit pool.
5. Teams ($30–$40 Standard / $120 Premium per seat): Adds SAML/OIDC SSO, centralized billing, team-wide privacy locks, and an internal marketplace for sharing custom .cursorrules, skills, and MCP plugins.