Gemini 3.5 Flash (Tested) Is Officially Better Than Claude & GPT

Added: 2026-05-20

[00:00:00]Hey guys, welcome back to another new exciting video. Finally, Google has published their new model which is the Gemini 3.5 flash and this model really have the great capability and also you can try this model inside anti-gravity and also inside Google Gemini app and this model are available for free to use. First, let me show you one of the output that I have got from this model.

[00:00:21]So, here you see I have made this website with this Gemini 3.5 flash and let me show you one of the advanced animation that is created by this model.

[00:00:29]So, if I open the website from beginning and if I scroll down below, here you see this kind of advanced animation that is made by 3.5 flash and if I scroll and here you see this kind of animation. Not only that, this one also and you know that when I tested the same prompt with other every model, let's say going 3.7 and also GLM MiniMax and Claude Sonnet Opus, I did not get this kind of advanced animation portion from any of the model.

[00:00:58]So, this is the first model who has made this kind of advanced animation and also the output is really better than the previous Gemini 3.1 Pro. So, the model is actually definitely worth it, okay?

[00:01:12]And also if I show you the comparison between the other models that they have given, first let me place myself on the right hand side. Okay, before that, let me show you one thing. So, in my previous going 3.7 video, one person commented that front end is mostly not coding, back end is for coding. So, I need you to test with models and tell. And also another comment that bro, you are testing only the front end. I need you to test with the back end, which model is best. All the models are different and best in specific case. See, the thing is that the line that you have written, now front end is mostly not coding and after that you also given the justification, front end can be done by anyone and now now not the back-end.

[00:02:00]See, you are writing this line means you are not capable enough to understand the front-end till now.

[00:02:07]So, please learn that. Learn all of the technology that is exist uh in front-end, React or Angular, and uh try to work on a real project. Okay, try to work on a real project that is uh industrial scalable project and uh you will learn there are a lot of things also there [snorts] in front-end, not only the back-end.

[00:02:30]Yes, I know back-end is also complex, uh but front-end is not that much easy. It is not about uh the landing page, okay?

[00:02:38]This landing page any model can make.

[00:02:40]And if you have watched my previous videos uh regarding uh Coin 3.7 or DC version 4 Pro and also Xiaomi Mi Movie uh version 2.5 Pro, then you know, right, that the output is not uh same from all of the models. The output uh can vary from different different models. And uh same thing for the back-end also, okay? If you are trying to work with a model, then the model should have the instruction following capability, should have the code uh logic building capability, right? And um yes, some models are good at front-end, some models are not. But the model that have the overall good capability, I actually prefer that model. And I am a big fan of Claude, okay? And I generally work with the Claude Sonnet and in very rare case I go with the Opus uh 4.6. But Sonnet 4.6 is my default favorite model. Why?

[00:03:35]Because the model have the good front-end capability and also good back-end capability. Now, if you want to try front-end for one model, back-end for one model, logic for one model, then you can go with that, okay? But I actually always prefer the one default model which have the average capability. Okay, means both of the capability should have. And I don't know that um which model you have tested, but the thing is that if a model can write the front-end logic in a very well manner, then I hope that the the model also will write the back-end code. And you know that there are a lot of videos on YouTube uh those who are reviewing this kind of new model, let's say Gemini 3.5 Flash or Queen 3.7. And are they making full end-to-end project? Full full-stack front-end plus back-end integration project? Most of the people are not doing that. They are simply reviewing the model and that is. But I am at least testing the model and I'm showing you the output. Okay. Now the total full testing you should do, right? If I show you all of that things and will you do will you not do anything?

[00:04:47]Okay, so please don't do this kind of uh comment that front-end can be done anyone. Okay. It is not correct.

[00:04:55]Okay, now let's back to this uh video.

[00:04:57]So uh if you see the benchmark on this coding on this agentic on this UI control, this model is giving the almost same kind of capability like that GPT 5.5. Okay. And also in some cases it is beating the Claude Opus 4.7. And all also the uh interesting thing is if you see this Claude Sonnet 4.6 and Gemini 3.5 Flash, almost in all areas it it has beaten this Sonnet 4.6. And also these things I am saying uh that this model 3.5 Flash have the capability between the Sonnet and Opus. And they will release the Gemini 3.5 Pro model in the coming weeks. So now imagine if a 3.5 Flash model have such a great capability, then 3.5 Pro also will have the more more advanced capability. So, I'm waiting for that. And if you want to try it, please go to this gemini.google.com and there you will find only this 3.5 flash model.

[00:05:57]And also they have released the high and low version. So, if you just go to this Let me show you. If you go to this anti-gravity and there you will find Gemini 3.5 flash high and Gemini 3.5 flash medium. Okay, both are fast. I have I have tried this. This model is too fast. Okay.

[00:06:19]Now, one thing is that the pricing. So, if you go to this Gemini 3.5 flash pricing page, as you see, there is a some free tier token usage also and also this paid tier. So, for paid tier, the input price is $1.5 and the output price is $9, which is huge than the previous 3.1 model, right?

[00:06:44]And also these are the some other pricing like contest casing, grounding with Google search.

[00:06:50]But the thing is that they have increased the 3.5 flash pricing, which is $9, which is more than the Sonnet, right? And also you see that it is justifiable because the output is better than the Sonnet.

[00:07:02]And if you go to this model details page, here you see it has the fast and smart means speed and scale don't have to come at the cost of intelligence.

[00:07:12]Means you will get both speed plus scale plus intelligence.

[00:07:16]And master the complexity deep reasoning across long horizon and interactive coding task and truly multimodal. It supports the text, audio, images, code and video. So, that's why they are charging more because you are getting the beyond Sonnet 4.5 capability. And also same as the GPT 5.5 with the flash.

[00:07:36]And this model have the knowledge cut off January 2025 and also it supports this function calling means tool calling structure outputs such as a tool and code execution. Okay, you can use it for the everyday coding task agent coding, advanced reasoning, multi-model understanding, long context understanding.

[00:07:52]Okay. And another thing is that I will make another separate video to compare this model with the GPT 5.5 and also all of the others like going 3.7 and also Xiaomi Mi version 2.5 Pro. I will compare the output and I will show you the real difference um >> [snorts] >> between this model on this on the same prompt I will try it, okay? And you will understand our real output difference.

[00:08:19]So, see you guys in the next video.

[00:08:20]Thanks for watching. Bye-bye. Take care.

[00:08:21]And if you have found this video helpful, don't forget to subscribe this channel. Don't forget to like this video also. See you guys in the next video.

[00:08:26]Thanks for watching. Bye-bye. Take care.