AI Assistant consuming credits much faster since WebStorm 2025.2.1

Hi,

Since updating to WebStorm 2025.2.1 (possibly also tied to the AI Assistant 2025.2 update), I’ve noticed a significant increase in credit consumption with my normal usage.

On older versions, it would usually take me around a week of regular work to use up about a quarter of my available credits (I’m on the Ultimate Tier). However, since the latest update, I used the AI Assistant for maybe two hours of typical usage and already hit a quarter of my credits.

Something seems off here, as this is a drastic difference compared to the previous builds. Could you please confirm if there were changes in how requests are being billed or if this is a bug?

Thanks!

36
53 comments

I see the same thing on my Junie consumption in phpStorm.

Last week I burned a whole month quota in intensive work all week.

Over 50 sessions with huge contexts, follow-ups and code/docs generated.

This week I topped-up the account for same amount as a month worth of credits and I depleted it in 3 sessions.

There is no way these 3 sessions were havier than 50 ones last week.

Same IDE, same Ai Assistant, I believe it did not update over last weekend.

This is definitely a bug, I have three other people complaining here

https://youtrack.jetbrains.com/issue/JUNIE-690

8

Hello,
 
Thank you for contacting JetBrains Technical Support.
 
We understand your concern, and you're right to notice a difference in quota usage after the recent update.
 
Starting from August 25th, we’ve adjusted the AI quota amounts across all plans. This change was made to align the quota more directly with each subscription tier — meaning you now receive an amount of quota proportional to what you’ve paid for.

This update is publicly documented in our sales article, with more context available in the "What’s changing" section:
https://blog.jetbrains.com/ai/2025/08/a-simpler-more-transparent-model-for-ai-quotas/
 
In practice, this does mean that quotas for some plans — including Ultimate — may be lower than before. If needed, you can purchase additional AI Credits to continue using the service after reaching your  quota.

To help manage your quota more efficiently, you can switch to a lower-cost model like Gemini Flash instead of GPT-5 when appropriate. Please also note that using Junie consumes more quota compared to other models.

For more information on why the cloud quota is being used up more quickly than before, please refer to this link:
https://youtrack.jetbrains.com/articles/SUPPORT-A-2539/Why-is-the-cloud-quota-consumed-faster-than-before

For guidance on what may lead to higher quota consumption:
https://youtrack.jetbrains.com/articles/SUPPORT-A-2448/What-can-lead-to-the-high-quota-consumption
 
Warm regards,
JetBrains Support Team

-22

Hello,

In my case this changes does not apply. I started using AI Assistant and in particular Junie more intensively since 25, so after your update.

It was fine, I spent all quota by friday in heavy load work.

IN 1 September, next monday, I paid for top-up same amount as the montly subscription (30 EURO).

I got only three sessions with that.

First week I had over 50 sessions similar like the three from monday, or even heavier.

Something happened over the weekend or top-up quota is messed up.

I saw the progress bar only about 1/5 loaded after top up.

 

There ar eother users noticing that, and a bug report submitted on youtrack

https://youtrack.jetbrains.com/issue/JUNIE-690/Junie-Ultimate-quota-calculations-seems-off

Thank  you

Cristian

4

I am having the same issue as the OP. My burn rate as at least 30x after the change. The quota would normally last me about a month. Now it may be gone after a day of use.  I'm also on the ultimate plan and it seems unreasonable that the my cost may go up 30x to maintain the same use I had only last week. 

12

Elina Burkhanova Hi Elina, thanks for sharing the resources. However, the main reason I upgraded was to avoid burning through my quota with general and junie usage. With this new change, it feels like Ultimate plan users who are already paying nearly £30 a month, are left with something that can be depleted in just a few days. That’s not why I chose the higher tier. To be honest, it comes across as a bit of a money grab, especially now that the option to purchase additional credits has been introduced. If this is to stick around, I'm likely to opt for third-part AI providers unless they can reintroduce the same quota as what was first set when I purchased the Ultimate plan.

11

I'm also an Ultimate subscriber and I'm also not happy with the changes you've implemented. It's draining too fast and I feel maybe Cursor is cheaper campared. Like it is, it'll be impratical for me at least. 

13

Not sure why Emilio Veloso and Liamlangstaff got downvoted. Honestly, I don’t see the point of using AI Assistant with these changes. I even upgraded from AI Pro to AI Ultimate and paid for the yearly plan.

I really hope they at least stick to their promise that yearly subscribers keep AI Ultimate the way it was until their subscription runs out. Otherwise, it feels like a bait-and-switch.

9

I’m also on the Ultimate yearly plan, and today is literally the first day I switched to this new subscription. What shocked me is that I’ve barely used the AI Assistant at all, yet almost half of my quota is already gone.

Honestly, this feels extremely disappointing. The way the quota drains now makes the whole yearly upgrade look like a bait-and-switch tactic — as if the promotion was just there to push me into paying for a full year, only for the service to become far less usable right after.

Right now, the so-called “monthly limit” is so restrictive that it doesn’t even compare to what I can do in just two days with Claude. That’s absurd. I regret upgrading to the yearly plan and I’m starting to lose trust in this platform.

I strongly hope JetBrains can recalculate quotas to a reasonable level for Ultimate members. Otherwise, what’s the point of paying for the top-tier plan if it doesn’t even sustain normal professional usage?

17

I use to get by with the AI Pro plan: most of the time I'd have enough quota to see me trough the month (I'm predominantly using AI chat, tried out Junie out of curiosity only).

August was the first time I saw a jump in credit consumption, but that was about the time I gave Junie a couple of queries so I figured it was related to that. 

Come September, using AI chat only, I burned trough my credits in around 3 days. I added 10€ worth of credits, disabled every possible setting, stopped allowing it to use access to codebase and manually included files, but I still burned trough that in around 3 days. I would say all those changes in the default settings seemed like I slowed the burn rate down a bit, but it's still nowhere near as slow as it was prior to August. 

I understand it was only a matter of time when the pricing for AI features would be risen, as is with any service that becomes popular, but this burn rate would effectively mean I'd need a plan of around 80€/month, given my usage, and I don't consider myself a hardcore user (e.g. not using Junie to generate heaps of code). 

Given that I also try and utilize a free plan of ChatGPT for general architecture debates, had I asked everything via AI Chat, I suspect I'd need an ever bigger plan given the burn rate.

The pricing of 80€/month (even if I upgrade to AI Ultimate, I'd still have top up for around 50€ each month), while debatably not that high of an expense, still seems a bit too much in terms of pricing I am seeing with some competitors, so I guess it's back to evaluating alternative solutions for my use case. It's usually more expensive to pay for everything as you go, so the existing AI plans, given the new burn rate, don't even make sense to me anymore.
 

13

I understand that you need to realign and review quotas, but for $30 USD, it really shouldn’t be used up within just 2 days. This just doesn’t feel normal. If someone subscribes to an Ultimate plan, you should expect that our demand in tokens (or whatever the metric is) will be high. I didn’t take this plan to ask for recipes or casual chit-chat, but to make the AI work hard. At this point, I’ll definitely start looking into alternatives, because the service feels far too resource-hungry for the price.

12

This needs to be addressed, can barely do a few days of work before hitting the quota on the ultimate plan…

4

There is definitely a real problem with this, my entire month's quota has been consumed in one day. I understand that there has been a readjustment of the usage rates, but according to JetBrains I have consumed $10 of Claude 4 Sonnet usage in a matter of 2 hours working on 4 files from a small project that have no more than 80 lines of code each.

It doesn't make any sense.

11

I'm testing using the AI Assistant in offline mode and connecting to xai/grok-code-fast-1 through LiteLLM proxy. Also the start of my rules .mdc file is like: 

You are an Angular 20+, Tailwind and NX expert. Only output *code, folder structures, or checklists*.
Do not provide any explanations, commentary, or reasoning.
For chat, format output exactly as requested, minimal tokens, ready to copy.

Uses very little tokens this way. I'm also trying to provide the context files directly to minimize search. I'll downgrade to Pro and use like this.

1

I purchased Ultimate and now this happened shortly after. Ridiculously inexcusable! Not going to be renewing and I may just decide to cancel my full JetBrains subscription as a result.

2

Paid for a year but this change just made the assistant not usable how can I get my money back?

4

Same here, ultimate subscription, I'm using AI Assistant at the same pace of last month.
The quota for last month was enough for the whole month, now I finished in 2 days

This is ridiculous. I'm not gonna pay 30$ for few queries a month.

6

Same shii..

It is not fair, Im not millionaire to spend a lot of money, even I havent job ,I want learn programming with AI Ultimate and now they steel my money, today I bought 10 euros and I havent them more . F.. ..it is robbery. If you  dont know how to get money ,dont use such way, U will loose more than will get . I studied at school where wasnt quality and now I want my personal teacher as AI and I cant use more it , I pay every month 30 for AI Ultimate, pro is useless for quantity of tasks I need to resolve, even Ultimate terminated with quotas much before month was finished. Im f..very unhappy with this sh..It is unfair to get more money from users. Maybe some can give idea better solution for Ai in other plaforms , local ollama same is uselless ,taking a lot of memory and impossible slow as snail.

5

I ran into the same issue when a process hangs in Junie’s terminal, there doesn’t seem to be a way to send a proper Ctrl+C like in a normal terminal. For now, the only workaround I found is to close/restart the terminal tab, but that feels a bit clunky.

This post was last modified: 19-09-2025, 23:15 pm by space waves

0

Mine has also drained very quickly. I am on ultimate. Where can I at least buy more tokens?

-2

Mavetyan dont't top up anyting. I wasted 30 EURO on that and got 2 sessions out of it. Something is totally off on this atm.
See here https://youtrack.jetbrains.com/issue/JUNIE-690

3

I don't understand the AI Assistant credit system. GitHub Copilot costs $10 a month to use the chat, but the 10 credits you get with AI Assistant don't even last a week without using Junie. It's absurd. I'm a fan of Jetbrains products, but I don't think it's normal for credits to run out so quickly. A couple of days ago, I ran out of AI Pro credits, so I bought 5 more credits, and they ran out in a couple of days. This is supposed to be half of what you get per month, and it's ridiculous that they run out in two days, and that's just for using the chat. With that expense between AI Pro and buying credits, I'd rather subscribe to Cursor and Windsurf, and I'm sure it would last more than a week. I think they should review how consumption is calculated, because it doesn't make any sense.

4

Same here. Just cancelled my subscription. I spend $20 in two days whereas I can work with Cursor a full month for that amount.

I do not want to spend too much time in configuring each prompt request, like turning codebase on/off, deciding if I should use “quick edit” instead of “Claude Agent”, or using “Chat” to generate to the code and copy it over, only because to save credits.

2

I'm reading posts from people talking about the Ultimate annual subscription. What's the difference, really? Both the monthly and annual subscriptions come with 35 credits per month. Those 35 credits are used up in 2-3 days of active use. So what? On top of those $300, I have to spend another $1,000+? Plus tax. Is everything okay with you? "ChatGPT uses a lot of credits, others use less." What the heck? If you buy a GPT Plus subscription directly from OpenAI, you can write an unimaginable amount of code in a single chat, and they won't tell you, "Take a month off without access, your limit is exhausted." Okay, Junie or the new Claude Agent, but a regular chat, generating commits, also requires credits? Well, this is just robbery... I really don't want to leave the wonderful JetBrains products, but it seems like I don't really have a choice...

5

This is a bait and switch.  You cannot change terms after someone paid you and subscribed to you service. I will definitely not be renewing my subscription. 

2

In the past half year I managed to exceed the Pro license only twice, where I had excessive coding marathons. Nonetheless I upgraded to AI Ultimate to have my peace of mind. Also I wanted to support the whole AI Assistant thing, so I gladly spent extra money on a subscription I don't even max out just because I was excited about it.

After one of the last updates I hit the limit for the first time with the AI Ultimate plan. I thought that's a bug, it's not possible after just some recreational coding and continued. But it was not a bug, so I topped up 5 credits. They lasted the day. So I added 30 extra credits but only after I read up on all the advice to limit credit burn, using smallest context sizes & cheap models. Yet, my top up credits lasted just 7 days. So I'd project to be spending ~100 EUR per month with considerate usage still. Needless to say that that's ridiculously much just for the convenience of having the chat inside the IDE.

So not only the extra cost is painful, but the side-tracking of “oh wait, I need to set the cheaper model for this simple question” and the agony of “oh crap, I'm still using Claude … ABORT ABORT!” are really not what a productivity tool that is literally meant to assist me should be doing. What was a real joy to work with is now a dreadful experience.

What's worse is that you can't even really control the cost. The whole idea of the AI assistant is to be magic. Magically adding all the context it requires for sophisticated answers and help. I have no idea how much the next chat prompt will cost me. I have no idea why my credits burn up so fast. There is absolutely no transparency other than keeping a very close eye on the credit counter and estimate how much I just wasted using the AI assistant while I could have just had the same conversation on my EUR 20/month GPT subscription I also have. Really, copy/paste to a browser window is less of a burden right now. A super dark shadow on an otherwise excellent product.

I'm curious if there will be any consolidation on this one. If not I'm cancelling the subscription. I have the All Products Pack. That little AI I'd really need inside the IDE I'd get there too.

3

Oh wow, that's really good to know! I use AI Assistant a lot too, so I'll keep an eye on my credit usage after updating. Thanks for the heads-up about WebStorm 2025.2.1!

1

I’ve been using AI mainly for documentation, unit tests. coding and error finding, and up until recently, I never had to worry about topping up credits. But over the past month, my balance has been draining within just a few days.

I really love JetBrains’ tools—they’ve been a huge part of my workflow—but it might be time for me to look into building my own AI rig with something like LM Studio and self-hosting models. It won’t be Claude, of course, but Qwen is getting pretty good—and at least I wouldn’t have to stress about credit consumption.

JetBrains, if you’re listening, please reconsider your usage policies. It’s becoming difficult to justify the costs, especially as competitors start offering more balanced and transparent usage models. I truly hope you find a way to make this more sustainable for your loyal users.

1

I canceled GitHub Copilot because I thought Junie was a better option for my work. I was foolish to sign up for the annual AI Pro subscription. From what I've seen in the comments here, even Ultimate has the same credit consumption. In the end, what you pay for is the plugin and a glimpse of AI after these changes. If I could, I'd ask for my money back, but I signed up in May, believing in the product they were offering. They created a lure and then showed the real product. I understand that it has to be profitable for both parties, but in the case of Junie, it's simply not worth using.  

4

I have also noticed that my credits have been drianing extremely fast this last month, before in general usage I would not even use half my monthly credits now I have already used 85% of my plans quota… this seems un-reasonable and I too will have to look into alternatives if this isn't addressed…

4

I have just asked the AI a question and it ended up including 25 versions of the same files repeatedly and has used up all of my remaining credits for the month and didn't even attempt to answer… You really need look into how the AI assistant optimises its file dependacy inclusions and put some limits on it… not impressed at all…

4

Please sign in to leave a comment.