Hey everyone,Due to the migration away from the old gemini-cli to the new Antigravity system, free tier users have been running into aggressive rate limits (429 Rate Limit Exceeded) extremely rapidly. With Antigravity being a multi-agent system that launches multiple subagents and reads from context blocks, the default model (High Tier) uses up the 1 million tokens per minute limit almost instantly.A lot of common CLI and application operations such as parsing simple log errors, terminal piping, and boilerplate syntax refactoring don’t require full reasoning capabilities.Suggested solution:Create an alternative model for Antigravity system called Gemini 3.5 (Flash Minimal/Lite), which would be less complex in parameters but have significantly looser or no token rate limits.
In that case call Gemini 3.5 Flash (low) with /fast. Gemini so far only load subagent if you explicitly tell them to do it or via skill /browser and /goal (or any skill that in their content states it).
nobody is a free tier developer xd bcs you dont develop anything at free tier think abt paying smth or getting in a little corparation where i am that pay everything you need for ur project 
bcs free thing are so bad
I mean yea, but still free people should be treated more too and even all paid users too, they all have unfair advantage with all of these bugs that drain their usage and google pushed it too far by making it do multiple tasks in the background and spawning sub agents, etc to explicitly drain their usage so fast like free users like me like when I give him a simple thing to do like fix a small bug it does fix anything because it spents time reading all the files and spawning sub agents, etc it drains all my credits doing absolutely nothing and now I have to wait 7 days until it does the same AGAIN, same for plus and pro and ultra users they also have unfair usages. They should actually add Gemini 3.5 flash (low) for just 1 month at least because of new AntiGravity app and cli.
I see the Free tier like other Free tier Google Services (think of Google Cloud related Services). They are nice to have so users can get their feet wet, make demos or hobby work, but for real work (read as something that makes you money back) then they should consider paying for the service, it cost them money to run it.
Like I mentioned before, Gemini 3.5 doesn’t spawn subagent unless instructed, directly or indirectly via skills (explicit or triggers) and rules. If it is happening, maybe is a bug 
. I remember having a similar issue with the old harness were Antigravity would auto-approve implementation plans and execute them when explicitly told to not do it.
Related to Free vs Paid tiers, IIRC, Plus got a 3x increase, Pro and Ultra got a 9x quota increase, maybe that’s why you may have that sentiment toward the Free tier quota. Now thinking about it, quota wise, is really bad …
| Tier Name | Price (USD) | Base Multiplier (Relative to Pro Plan) | 3x Quota Upgrades Applied | Total Compounding Boost (Post-Backlash) | Target Audience / Features |
|---|---|---|---|---|---|
| Google AI Free | $0/mo | N/A | No | Meaningful quota, refreshed weekly Weekly rate limit * | Casual exploration |
| Google AI Plus | $10/mo | Fractional | Yes | 3x multiplier on lower base | Entry-level coding |
| Google AI Pro | $20/mo | 1x Baseline | Yes | 9x total boost (3x, then 3x again) | Standard developers |
| Google AI Ultra ($100) | $100/mo | 5x more token quota | Yes | 45x total boost (5x base × 9x upgrade multiplier) | Power users & startups |
| Google AI Ultra ($200) | $200/mo | 20x more token quota | Yes | 180x total boost (20x base × 9x upgrade multiplier) | Full-scale enterprises |
*: I cannot get exact data about this one, at some point it was a 250 agent requests per day limit, then it got reduced to 20 daily requests. Quota reduction numbers make sense though.