I ran an experiment: I asked Gemini for a plan to make a change to my app. Then, I passed that instruction to Antigravity and Cursor. I fed both of those plans back to Gemini for a verdict: Cursor’s was MUCH better. Cursor’s free plan includes 200k tokens; it used 67k. Antigravity shows me this.
Oh, and those two queries to Gemini? They consumed 49% of my quota (Pro plan).
Congratulations Google, you killed two birds with one stone.
Shared Quota Across Gemini Models
Until now, users have had separate rate limits for Gemini Flash and Gemini Pro models. Now, we are combining these into a single rate limit, drawn down as per API pricing, allowing you to get more of whichever model you prefer. So, for example, if Gemini Flash is 8x cheaper than Gemini Pro as per API pricing, you will be able to use X tokens on Gemini Pro, 8X tokens on Gemini Flash (provided the same ratio of input, output, and cache read tokens), or any linear combination in between. Previously, we essentially forced a particular linear combination by setting independent rate limits, but now it is one shared pool.
What kind of pun is this…
