Hey folks — been using Tabnine in WebStorm for a few months now. Love the completions, but lately I’ve noticed my IDE gets really sluggish, especially on larger projects.
Anyone else seeing this? For me, it’s:
- Slow file indexing on startup
- Typing lag when Tabnine is suggesting
- High CPU usage when it’s active
I’m on the latest versions of both. Already tried disabling other plugins, but it still feels heavy. Any tips or config tweaks to speed things up?
— Jason
Let's build better workflows.
I've run into similar performance degradation with Tabnine, specifically around the typing lag and high CPU. The issue isn't just Tabnine itself, but how it interacts with WebStorm's own indexing and code analysis on larger codebases.
What helped in my case was adjusting Tabnine's context configuration. By default, it tries to analyze a massive amount of project context, which for a large project creates significant overhead. Go to Tabnine's settings in WebStorm and reduce the "Maximum Number of Results" and the "Maximum Context Length" (if visible). This limits the scope of its model inference.
Also, check if you're using the cloud-based model versus a local one. The cloud model adds network latency on top of processing, which can feel like sluggishness. Switching to a smaller, local model might trade off some completion accuracy for significantly lower latency and CPU load.
Monitor your CPU usage in the activity monitor while WebStorm is idle but open. If Tabnine's daemon process is still consuming high single-core CPU, it's likely stuck in a background analysis loop. Sometimes a full restart of the IDE is the only temporary fix, which points to a memory leak in the plugin.
Show me the numbers, not the roadmap.