Sweep AI

Sweep AI is the #1 rated AI coding assistant for JetBrains IDEs, built specifically for developers who live in IntelliJ IDEA, PyCharm, Android Studio, WebStorm, PhpStorm, Rider, and the broader JetBrains ecosystem. Founded in 2023 by William Zeng (CEO) and Kevin Lu — both former engineers at Roblox — Sweep launched at Y Combinator Summer 2023 with an original vision of an AI junior developer that turned GitHub issues into pull requests. After raising $2.5M (YC + $2M seed from Goat Capital, Replit CEO Amjad Masad, and Exceptional Capital at a $25M valuation), the team pivoted to building the AI coding assistant they personally wanted as daily JetBrains users, reaching $1M ARR by June 2024 with just 5 people.

Sweep's plugin has three main capabilities. Sweep Tab is a next-edit autocomplete powered by a custom 1.5 billion parameter model (Sweep Next-Edit) trained specifically for the JetBrains edit pattern — it suggests precise multi-line code changes at millisecond latency. The Sweep Agent supports multi-step coding tasks using a choice of models including Sweep's proprietary Sweep-0.1 model, Claude Opus 4.5, GPT-5.2, Gemini 3 Pro, GLM 4.6, and Grok Code Fast 1, with full codebase indexing, MCP server integrations (with full OAuth 2.0/2.1), web search, bash tooling, and AI code review. Inline editing, AI commit messages, and configurable custom prompts round out the feature set.

Sweep runs its own LLM infrastructure for unmatched price, performance, and privacy — no code is retained by Sweep or third-party providers after a session (zero data retention by default). The platform is SOC 2 compliant, supports BYOK for full control over API routing, and offers team-wide Privacy Mode enforcement for enterprise users. With 4.9 stars and 40,000+ installs, Sweep is trusted by developers at Ramp, Amplitude, Atlassian, Klook, INDmoney, and the MTA, who call it "a million times better than JetBrains' own tools" and "the best AI coding assistant" they've tried including Augment Code and Windsurf.

Sweep uses a usage-based credit model. The free tier gives limited Sweep API Credits for testing all features. All paid individual plans (starting at $20/month) grant the plan cost as included API credits plus unlimited Tab autocomplete suggestions. Unused credits roll over for 30 days.

Agent usage is charged at the underlying LLM provider cost plus a 5% Sweep markup: Sweep-0.1 (fastest, ~$0.02/message), Claude Opus 4.5 or GPT-5.2 (~$0.25/message), Sonnet 4.5 (~$0.15/message), Gemini 3 Flash (~$0.035/message), and GLM 4.6 (~$0.028/message). Sweep Next-Edit Tab autocomplete is included at no per-message cost on all paid plans.

Team plans include $40 in Sweep API credits per seat with pay-as-you-go overages billed to the organization. Team plans add SSO, team-wide usage visibility, and the ability to enforce Privacy Mode org-wide. Contact team@sweep.dev for volume discounts on larger teams. BYOK (bring your own key) is available to route all model calls through your own API accounts at zero markup.