Our new Gemini 3.5 Flash is our strongest agentic and coding model yet, delivering frontier-level performance at 4x the speed of comparable frontier models at less than half the cost. We had Tulsee Doshi break down what you need to know about it.

@tulseedoshi 4x the speed and 6x the cost. (Twice the tokens and 300% price increase) Not sure it'll be all that well received.

@tulseedoshi Limits limits limits lim

@tulseedoshi What's the point using it when usage limits are used up with just a few prompts

@tulseedoshi And we can't use it with these horrible usage limits! So who is that for? Google still silence about the that change. Horrible way to treat your subscribers 🫡

4x speed at half the cost only matters if the agentic reliability holds up under real workloads. The bottleneck isn't inference throughput—it's how often the model chains itself into dead ends during multi-step tasks. Would love to see eval numbers on task completion rate, not just benchmarks.

@tulseedoshi Oh and the massive 6x token burn, inability to code, and insane unusable usage caps? Are you breaking that down too?

@tulseedoshi Hardly

@tulseedoshi Alright I’m tired of this. Not only is it not smart it’s also costly. I don’t understand Fast and dumb and costly why are you guys releasing worse products. Seems like Google is turning into the next Microsoft. Copy others and release a bad product instead of something better

@tulseedoshi وصولٌ يختصر المسافات بين الفكرة وتحقيقها يا صديقي. السرعة والكفاءة هنا ليستا مجرد أرقام، بل هما وسيلة لتوسيع آفاق الإبداع وجعل المعرفة أكثر قرباً من الجميع. العلم يزدهر كلما منحنا الوقت لنفكر ونبدع أكثر. ✨🌱

@tulseedoshi It is looking like speed/cost will matter more than the leaderboard line. Interesting to see how this plays out. If agents are going to run in the background, iterate, inspect, and recover, latency becomes part of product quality.

@tulseedoshi Frontier-level at 4x speed for half the cost is the real story here — agentic loops only work when latency + cost don't kill the unit economics. Unpacking the implications on @techimpactTV this week.

@tulseedoshi 4x the speed at half the cost. The AI price war just got very real 📉🔥

@tulseedoshi It’s all about speed, cost, and coding quality now. Very impressive release. If this holds in practice, Gemini 3.5 Flash will be a truly exceptional coding partner.

@GeminiApp @tulseedoshi It thinks what to reply to “Hi”… to be fast it needs to have good adaptive thinking

@tulseedoshi Try to do something with the Pro subscription now... 🤣🤣🤣 Fuck you Google!!! You lost another customer.

@GeminiApp @tulseedoshi I dont think speed is a good thing to brag out. Unlike cost per intelligence

@GeminiApp @tulseedoshi Usage limits sucks, Google. 🤡

@tulseedoshi This is nice case of marketing trying to save completely bad product. Just say the same thing again and again and hope that people will believe it no matter how to product works in real situations.

@tulseedoshi Google Korea B2C "ME"

@tulseedoshi Excited to put Gemini 3.5 Flash to the test!

@tulseedoshi Using Gemini 3.5 in production now. The video capabilities are incredible. Built a tool that auto-summarizes saved content ,

@tulseedoshi

@tulseedoshi Speed changes everything. When AI is fast enough, you stop treating it like a separate app and start using it in the middle of thinking and doing.

@tulseedoshi This model demonstrates how cutting edge AI can deliver actionable results faster than ever. Exciting times for coding and agentic AI applications.

@tulseedoshi Let's talk about the price 🙂

@tulseedoshi News from Google

@tulseedoshi That's good 👍

@tulseedoshi This feels like a real shift toward practical agentic AI. Curious how it actually handles complex real world workflows at scale.

@tulseedoshi Unethical Google is turning the internet into a wealth and value transfer for themselves. Here's some alternatives

@tulseedoshi 不是你说的这样

@tulseedoshi 4x speed at half the cost and it's still topping agent benchmarks. Google is not playing

@tulseedoshi Nói gì đi @grok