JetBrains shipped a meaningful update to their AI Assistant this week. The headline changes: Claude 3.5 Sonnet as a model option, a free tier for individual developers, and an in-IDE agent mode that approaches what Cursor’s Composer offers.
For developers loyal to IntelliJ, PyCharm, GoLand, RubyMine, etc., this changes the AI tooling calculus. You no longer have to pick between “the IDE you know” and “the AI features Cursor has.”
What’s new
The update includes:
Claude 3.5 Sonnet support. Previously, the assistant was OpenAI-backed. Now you can pick between OpenAI models, Anthropic models (Claude 3.5 Sonnet, Claude 3.5 Haiku), and Google models (Gemini 1.5 Pro/Flash).
Free tier. Individual developers get a limited free tier — about 100 messages/month, basic code completion. Heavy users still need a paid subscription. The free tier closes the price gap with Cursor’s free tier.
Junie (agent mode). JetBrains’ agent product, “Junie,” is now in public beta. Junie operates similarly to Cursor’s Composer — given a task description, it makes changes across multiple files, runs tests, and reports back. Currently available in IntelliJ IDEA, PyCharm, and WebStorm.
Improved code completion. The inline completion model has been upgraded. Latency is comparable to Copilot; quality on JetBrains-specific languages (Java, Kotlin, Ruby for RubyMine users) is competitive.
Why this matters
JetBrains’ AI story has been weaker than competitors for a while. The AI Assistant launched in 2023 was usable but felt like a port of an external feature, not a native part of the IDE. Many developers in JetBrains ecosystems used Copilot or Cursor instead, and JetBrains was losing ground.
This update closes most of the gap. The features now match or exceed Cursor for most workflows. The integration is deeper because JetBrains owns the IDE.
For developers who’d been waiting for “Cursor for IntelliJ,” that’s roughly what JetBrains has shipped.
Where it’s still behind
A few areas where Cursor or other tools still lead:
Codebase indexing. JetBrains’ indexing is good for IDE features (navigation, refactoring) but the AI doesn’t fully use it for chat queries. Cursor’s chat-side indexing is more sophisticated.
Composer-style multi-file editing. Junie works for autonomous tasks but the human-in-the-loop multi-file editing experience (Cursor’s Composer) is more polished. Junie is more “kick off and walk away”; Cursor offers both modes.
Per-language defaults. JetBrains’ assistants are uniform across languages. Cursor’s per-rules system feels more tailored. Both can be customized; JetBrains’ configurability is less prominent.
Pricing transparency. Cursor’s pricing tiers are well-understood ($20 Pro, $40 Business, etc.). JetBrains’ AI Assistant pricing is a bundle (included in some IntelliJ Ultimate subscriptions, separate purchase otherwise) that’s harder to comparison-shop.
What’s interesting
A few aspects of the release worth noting:
Junie’s terminal integration. Junie can run commands and observe output, similar to Claude Code. This bridges the agent and CLI categories that have been separate. JetBrains is positioning their IDE as the integration point for agentic work.
Multi-language model selection. Different languages can default to different models. For example, you could have Claude for Python files and GPT-4o for Java files. The configuration is per-project.
Privacy controls. JetBrains’ enterprise tier has stronger data residency commitments than Cursor’s. For organizations with compliance requirements, this is a real differentiator.
Comparison shopping
If you’re a JetBrains user evaluating AI tools:
- Stay on JetBrains: the new features close most of the gap with Cursor
- Switch to Cursor: only if specific Cursor features (Composer specifically) are essential
- Try both: the cost is bounded; switching back is free
If you’re a Cursor user considering JetBrains: the IDE features differ enough that “switch to evaluate” is a real exercise. JetBrains’ refactoring tools, debugger, and database integration are stronger than VS Code/Cursor’s. The tradeoff is the AI features being slightly less polished.
For Java/Kotlin developers especially, JetBrains’ updated AI removes the main reason to consider Cursor.
What’s next
JetBrains has hinted at:
- Junie expanding to more IDE products (RubyMine, GoLand, etc.)
- Better integration between AI Assistant and the IDE’s own intelligence (refactor suggestions powered by both)
- An MCP-equivalent for connecting external tools
Their roadmap has been less aggressive than the AI-native tools, but the pace is picking up. Six months from now, the gap may narrow further.
The competitive landscape after this release
The categories of AI coding tools as of now:
AI-native IDEs (Cursor, Zed, Windsurf): built around AI, fork or build editors. Strong AI integration. Smaller existing user bases.
Established IDEs with AI add-ons (JetBrains, VS Code + Copilot, Eclipse): existing IDEs adding AI. Improving rapidly. Larger user bases.
CLI agents (Claude Code, Aider, Codex CLI): terminal-shaped. Different workflow. Growing user base.
Editor extensions (Cline, Continue): add agent capabilities to existing editors. Workable but not as polished.
The categories overlap. Many users use one primary tool plus one secondary. The choice depends on workflow, tolerance for new tools, and feature priorities.
JetBrains’ update strengthens the “established IDE + AI add-on” category. For some developers, this is the path of least resistance — they keep their familiar IDE and get AI capabilities. For others, the AI-native tools’ deeper integration is worth the migration cost.
Worth trying?
For JetBrains users who haven’t actively evaluated AI tools yet: yes. The free tier removes the cost barrier. The features are competitive. There’s no excuse not to evaluate.
For developers committed to Cursor or other AI-native tools: probably skip unless you have other reasons to use a JetBrains IDE. The features overlap; the migration cost isn’t justified by the AI improvements alone.
For Cursor evaluators considering JetBrains: try both for a week each. The IDE differences will likely matter more than the AI differences for your decision.
The market for AI tooling is now mature enough that switching costs are real but bounded. The AI Assistant update gives JetBrains users a credible path forward without leaving the ecosystem they’ve invested in. That’s a meaningful improvement.