AI Pro Subscription

Google AI Studio
123

The same problem, and unstable.

124

I’ve been actively using Google AI Studio Playground again for about a week now. I have a Pro subscription, and Gemini 3.1 Pro in my ~500k-token chat only lets me make around 20 requests per day.

However, Gemini 3.8 Flash, 3.7 Flash, 3.6 Flash, and 3.5 Flash don’t seem to work with a context this large for some reason. The only Flash model that works is Gemini 3 Flash, but even then, I’m limited to around 15–20 requests per day in this chat.

I had the exact same issue back in early September, and apparently nothing has changed since then. The only difference is that Gemini 2.5 Pro was also available back then, so I could alternate between 3.1 Pro and 2.5 Pro. Now I’m basically forced to use 3.1 Pro only.

125

My experience today. A long chat, about 420,000 tokens. Gemini 3.0 Flash processed 20, maybe 25 requests in it—I didn’t count exactly—and then hit the limit. Naturally, Gemini 3.0 Flash continued to work normally in new chats. I’m still hoping this is a bug and not intended behavior.

By the way, does anyone have an account without a subscription to check what the limits are without one?

126

So, here’s what happened on October 7. After about 25–30 generations with Gemini 3.0 Flash in a long conversation with a large context, it stopped responding in long conversations (reached the limit), but continued to respond in new ones.

After about 10-11 generations with 3.1 Pro (of which about 7-8 were in a conversation with 890,000 tokens of context, 2 in a conversation with 300,000 tokens of context, and 1 in a conversation with 11,000 tokens of context), 3.1 Pro hit the limit in absolutely all chats.

Honestly, these limit values remind me more of the Free Tier without a subscription…

Update: In Gemini 3.0 Flash, the chat limit of 890,000 tokens was reached with the second-to-last message. And now, an hour and a half before midnight Pacific Time, 3.1 Pro was able to generate a response again (only one so far; I haven’t tried again).

127

If this situation continues like this and gets worse and worse I will probably think about an alternative… today the problem of all of us still persists even more severe…

128

I guess this is our new reality now: reduced quotas, increasingly aggressive censorship, and an overall decline in the quality of a paid service. There are no answers, no explanations, and nothing is clear.

How are everyone’s quotas looking today?

129

I’ve been trying to keep my use of AI Studio to a minimum lately, so I can’t say for sure about the quota. But yes, the last time I hit the limit, it was lower than before.

And the “silence” tactic that Google appears to have adopted is much worse than if they had simply stated that this is intended behavior or acknowledged a bug.

I wonder if Google’s decision to disable the Gemini 3.5 Flash and 3.7 Flash models will affect the limits in any way?

130

What alternatives to Google AI Studio are you personally considering? I think that if this nightmare with the limits is here to stay, I’ll switch to Claude Console (or Claude Platform—I don’t remember exactly what their platform with the API and Playground equivalent is called) or OpenRouter. What do you think? Maybe you could add to the list?

131

Still cooked on my end. Around 10 prompts daily as an AI Pro subscriber, and no response whatsoever from these people. They just don’t seem to care.

132

Well… Two models have been disabled, but the limits for the rest haven’t changed at all. About 20 generations with Gemini 3 Flash in a long dialogue totaling 430,000 tokens today

133

It is really disappointing because my AI PRO plan has finished and with Gemini 3.1 pro I had about 5 6 requests and well it is very close to your AI PRO quota which is 10 and it seems that there is no difference between having a subscription or not having a subscription

134

Hi everyone! In my chat with a context of approximately 600,000 tokens, I’m now limited to just 15–16 requests per day, not even 20 anymore.

It’s really frustrating because I’m running a long, literary-style roleplay that only Gemini Pro seems capable of handling properly in terms of character consistency and remembering past events. I don’t particularly enjoy using other AI models or services with dedicated roleplay bots, either.

And 16 responses a day really isn’t much at all. I have to cut back on the actual roleplay interactions quite drastically, which is gradually turning the experience into something closer to writing a novel than having an ongoing, interactive roleplay with the AI.

135

image
image596×164 7.84 KB

There’s also a new UI now that pops up, with them having pernamently pushed the “select a different model” troubleshooting step as a primary workaround. The product has become an absolute joke, and unfortunately, these downgrades feel like a fixture of it now. I’m sorry everyone;;

136

Maybe this is just new text for the limit errors? If not, then the following behavior seems strange:

  1. The Gemini 3 Flash model hit a limit in a dialogue with 213,175 output tokens and 221,010 outputs. As I understand it, when a request is sent, all 430k tokens are sent, and they are processed as input.

  2. However, when uploading a PDF containing 470k tokens in a new chat, Gemini 3 Flash generates a response normally.

Conclusions:

  1. Either these 470k tokens are actually different from the 430k tokens in the long dialogue,

  2. Or this is some kind of bug. Because if the tokens in the PDF are equivalent to regular ones, then once the token limit was reached, the model shouldn’t have processed the PDF and generated a response

In any case, I believe that if Google respects its users at all, it should explain in more detail exactly how the new limit system works.

137

Without a subscription, 5–6 requests to Gemini 3.1 Pro?

And how many requests does 3 Flash allow without a subscription?

138

Gemini 3 flash is a paid model and I cannot use it without ai pro and well gemini 3.1 flash lite after the finishing of the quota of the gemini 3.1 pro model I can give it only about 5 requests

139

Oh, I forgot that Flash 3.0 isn’t free anymore. What about 3.8 Flash?

140

Almost like 3.1 pro about 5 to 10 requests

141

So, Google has released Gemini 4 Argon in AI Studio (currently only for Google AI Ultra and the Paid API).

  1. If there are people here with AI Ultra or those using the Paid API, how good is Argon?

  2. Has anyone’s quota situation changed for the better or worse since Argon was released?

  3. Do you think they’ll release Argon for Google AI Pro subscribers, and if so, when? I’m guessing in about a month.

142

Exactly I also really like to know how Argon is and can change my opinion about buying ai pro or api again of course for the ai pro subscription it is not available?