Tabnine's pricing just went up - is it still competitive?

Hey folks, I was just updating my plugins and got the notification about Tabnine's new pricing structure. I've been a Pro subscriber for about a year now, leaning on it heavily for JavaScript/TypeScript and Python work in VS Code, and this increase has me doing a double-take.

The old $12/month (or $99/year) for Pro felt like a sweet spot. Now, the new "Starter" plan at $9/month seems positioned against GitHub Copilot's $10, but it's missing key features like the full-line completions and IDE function suggestions that I use constantly. To get those, you need the "Pro" plan, which is now **$19/month** or $189/year. That's a significant jump, nearly doubling the effective cost for the feature set I currently rely on.

So I'm left wondering: does this new positioning hold up under scrutiny? With competitors like:
- **GitHub Copilot** at $10/month (or free for some)
- **Codeium** and **Sourcegraph Cody** offering robust free tiers
- Even **AWS CodeWhisperer** being free for individual devs

...the value proposition feels like it's shifted. Tabnine's local model options and on-prem deployments are still a unique selling point for some enterprises, but for an individual developer or small team?

Here's a snippet of my typical usage pattern—Tabnine excels at this kind of stuff:

```javascript
// I start typing a common hook pattern...
const [data, setData] = useS
// And it would immediately suggest:
// State(initialState) // which is great.
```

But now I'm asking: is this fluency worth nearly $20 a month more than the competition? The completions are good, sometimes more "conservative" but also more predictable than Copilot's. I also really value the privacy-first approach.

Some specific points I'd love to get the community's take on:

* **For those who use the local models:** Does the performance and accuracy justify the price increase, especially compared to cloud-only alternatives?
* **Has anyone compared the "Starter" plan's limited completions to the full Pro experience?** Is it too gimped to be useful?
* **Where does Tabnine still have an edge?** For me, it's been in:
* Less "noisy" suggestions in well-defined codebases.
* Configurable granularity for completions (line vs. full function).
* The middleware model that can run locally.

I'm genuinely on the fence about renewing. The jump feels steep, but I also dread the productivity dip and retraining period if I switch. Curious to hear your workflows, comparisons, and whether you think the new pricing aligns with the value you're getting.