YouTube2Text

Google ha Vuelto! Lanzó GEMINI 4 Hydrogen al fin GRATIS y SIN Límites — Transcript

by Alejavi Rivera · 4,757 words · 704 segments · language en · Watch on YouTube

Full transcript

  1. 0:00Just because everyone thinks you’ve
  2. 0:01fallen behind doesn't mean you’ve lost
  3. 0:03the race.
  4. 0:05Google has just returned by launching
  5. 0:07its latest model, Gemini 4 Hydrogen. It
  6. 0:10performs similarly to the top models
  7. 0:12from its competitors, but with one clear
  8. 0:14difference. It is accessible to everyone
  9. 0:17for free. I’ve been testing it
  10. 0:19thoroughly this past week, and honestly
  11. 0:21, it’s the fastest AI that offers the
  12. 0:23best results.
  13. 0:29To prove it, we will put it to the test
  14. 0:31with use cases like generating any
  15. 0:33digital asset, 3D simulations, complex
  16. 0:35full-scale games, or even creating
  17. 0:37videos, among other use cases.
  18. 0:40The best part of all this is that beyond
  19. 0:42being a free model, even if they haven't
  20. 0:44publicized it, it could potentially be
  21. 0:46used unlimitedly. Sounds good? Well,
  22. 0:48let’s get to know all about it.
  23. 0:52Let’s start by getting to know a little
  24. 0:55better what this new Gemini 4 Hydrogen
  25. 0:56model consists of. The fact is, this is
  26. 0:59the replacement for the Gemini Flash
  27. 1:01models. Google reports that it is up to
  28. 1:042.5 times faster than its previous model
  29. 1:06, Gemini 3.8 Flash.
  30. 1:09It is better at programming, jumping
  31. 1:12from 18.2%to 57%on Terminal Bench 4.1.
  32. 1:16Artificial Analysis even positions its
  33. 1:18reasoning above the Argo model, which
  34. 1:20was only available as a paid service.
  35. 1:23It is a model that also contains a
  36. 1:24context window of over 1 million tokens
  37. 1:26, is multimodal—able to understand
  38. 1:28images, audio, video, documents, and
  39. 1:30text—and is optimized for all agentic
  40. 1:32and tool functionalities.
  41. 1:35But this is not just another model
  42. 1:37presented as an improvement on Google's
  43. 1:40line, as it is currently in the top five
  44. 1:42, above models like Faibol 5.1 or GPT
  45. 1:45Astra 6.
  46. 1:46Except this one is even faster and free
  47. 1:49, and its results are surprising.
  48. 1:52Take a look at this comparative Motion
  49. 1:54Graphics video where we have a superior
  50. 1:57model like the Gemini 4 Carbon and the
  51. 1:59new Gemini 4 Hydrogen, where we can see
  52. 2:01it did it in practically half the time
  53. 2:03and at less than half the cost.
  54. 2:06But what happens if we compare it to
  55. 2:07other models like Claude?
  56. 2:09Well, here with this animation of a
  57. 2:11floating house, we can see how Favol 5.2
  58. 2:14had a cost of about 19 dollars and took
  59. 2:16over 45 minutes to work, while if we go
  60. 2:19directly to this new free model from
  61. 2:21Google, it was the fastest by far,
  62. 2:23taking barely 12 minutes and costing
  63. 2:26approximately just 1 dollar.
  64. 2:29It looks very promising, so let’s
  65. 2:30discover where we can use it, and watch
  66. 2:32out, because its power changes a lot
  67. 2:34depending on where we are using it.
  68. 2:37We are going to start with the official
  69. 2:39Google Gemini platform, since now, even
  70. 2:41if you have a free version, like in my
  71. 2:43case with this account here which, by
  72. 2:45the way, will be the same free account
  73. 2:47we will be using throughout the entire
  74. 2:49tutorial, when we click on the model
  75. 2:51selection section, we no longer find the
  76. 2:533.8 Flash model, because even though we
  77. 2:55only see that we cannot use Gemini 4
  78. 2:57Argon, the truth is that by default all
  79. 2:59models are Gemini 4, meaning that now
  80. 3:01the free versions have those from
  81. 3:03Hydroge.
  82. 3:04From here we will also be able to switch
  83. 3:06between the level of reasoning, and this
  84. 3:08really changes the result a lot. Let's
  85. 3:10start by putting it to the test. For
  86. 3:12that, I will tell it here to simply
  87. 3:14create a professional website for an AI
  88. 3:16automation agency. I click send and
  89. 3:18notice that it starts programming right
  90. 3:21here, and if we go to the code section,
  91. 3:23we can see how all this code is really
  92. 3:25being generated very quickly.
  93. 3:28A little more than a minute later, we
  94. 3:30had the result here, and if I expand it
  95. 3:32, we can see how it has laid out this
  96. 3:34very futuristic webpage, fitting for
  97. 3:36that AI automation agency.
  98. 3:39From here we can see different blocks,
  99. 3:40and some calculators also appear that we
  100. 3:42can interact with, as you can see right
  101. 3:44here.
  102. 3:45We can see other sections, and if we go
  103. 3:47all the way to the bottom, we can see
  104. 3:49how it has actually completed the entire
  105. 3:51website, and most surprisingly, in
  106. 3:53record time. Let's put its physics to
  107. 3:56the test now.
  108. 3:57To do this, in a new chat, I will enter
  109. 3:59the following prompt asking for a 3D
  110. 4:01medieval siege game with physics.
  111. 4:03In short, a wall that appears so I can
  112. 4:05knock it down to see all the movements
  113. 4:07and how natural they are on the first
  114. 4:09try.
  115. 4:10To see this, I will click send again,
  116. 4:12and in just about a minute later, we
  117. 4:14have this simulator here.
  118. 4:16To see it better, I will expand it, and
  119. 4:18notice how we have this wall here; by
  120. 4:20zooming out a bit, we can see how we can
  121. 4:22launch a stone from here.
  122. 4:25So if I click fire, we can see we would
  123. 4:27have missed, so let's change the angle a
  124. 4:30bit lower, I'll add a heavier stone and
  125. 4:32reduce the power, and if I click fire,
  126. 4:34look now, it seems like it's going to
  127. 4:36crash, slow motion activates, and we can
  128. 4:39see how everything is actually working
  129. 4:41well with each of the movements.
  130. 4:44And all of this on the first try, but
  131. 4:46watch out because normally the effort
  132. 4:48level is at the minimum.
  133. 4:50Returning to where we left off, notice
  134. 4:52that if I start a new chat from here,
  135. 4:54when clicking on the reasoning level, we
  136. 4:55will normally find it at the fastest and
  137. 4:57most efficient setting, but by moving it
  138. 4:59to high, we will be able to get a
  139. 5:01somewhat better result.
  140. 5:03I'll give it the following prompt where
  141. 5:04I want it to create a 20-second vertical
  142. 5:06motion graphic style video that talks
  143. 5:07about fun facts about octopuses and why
  144. 5:09they have three hearts.
  145. 5:10If I leave the reasoning level at small
  146. 5:12here and click send, the first thing it
  147. 5:14will tell me is that if I want to
  148. 5:16generate it, I should switch to a higher
  149. 5:17version, but the truth is that this is
  150. 5:19something not necessary to be able to do
  151. 5:21it. Notice that if I now reply for it to
  152. 5:23do it in code or to do everything
  153. 5:25possible to make it happen and click
  154. 5:27send, then effectively, it would tell us
  155. 5:29here that it is already working on this
  156. 5:30video. A minute or so later, we can
  157. 5:32already find the video here. To see it
  158. 5:34better, I'm going to enlarge it. We have
  159. 5:36the video here now, so I'm going to
  160. 5:38click play to see if it appears
  161. 5:39correctly. Notice here that an error
  162. 5:41actually appears, so here we would have
  163. 5:43to iterate, and really, not only did it
  164. 5:45not give us the result on the first try
  165. 5:47, but it also did it with a small error
  166. 5:49.
  167. 5:50However, if I now switch the reasoning
  168. 5:52level to high in a new chat and give it
  169. 5:54the same prompt, notice that right away
  170. 5:55, if we click send, notice that on the
  171. 5:57first try, without having to urge it to
  172. 5:59make this video, it would immediately
  173. 6:01start working on it, and a little later
  174. 6:03, it would have delivered this other
  175. 6:05result on the first try.
  176. 6:07Let's watch those 20 seconds to see if
  177. 6:08it did it better.
  178. 6:10>> This animal has three hearts. Two of
  179. 6:14them pump blood directly to the gills to
  180. 6:16oxygenate it.
  181. 6:20And the third one pumps that blood to
  182. 6:22the rest of the body.
  183. 6:25When it swims, that heart stops beating
  184. 6:27. So
  185. 6:28>> And look, a result that is much better
  186. 6:30where we can get more out of this new
  187. 6:32Gemini model. With what we've just seen
  188. 6:34, we can create incredible content with
  189. 6:36artificial intelligence, but there is
  190. 6:38something that can completely change how
  191. 6:40we perceive it: sound.
  192. 6:42That is why I want to introduce you to
  193. 6:44Epidemic Sound, where we have more than
  194. 6:4655,000 songs and 250,000 sound effects.
  195. 6:49Furthermore, from here we can search for
  196. 6:51everything in a very easy way.
  197. 6:53If, for example, we start looking for
  198. 6:55music that fits the type of content we
  199. 6:56need, artificial intelligence will
  200. 6:58directly provide us with different
  201. 7:00recommendations that we can preview and
  202. 7:02download. We will also be able to do the
  203. 7:04same thing with sound effects.
  204. 7:05We just have to specify what we are
  205. 7:07looking for, for example, intensity in
  206. 7:09transitions or highlighting specific
  207. 7:10moments, and we would have all those
  208. 7:12sound effects right here to download and
  209. 7:14implement in our project.
  210. 7:16And there is something fundamental here
  211. 7:17that we must talk about: their license.
  212. 7:19A subscription allows us to use their
  213. 7:21catalog in our content and platforms
  214. 7:23without having to worry about copyright
  215. 7:25claims.
  216. 7:26In addition, videos published during an
  217. 7:28active subscription keep their license
  218. 7:29even after you cancel it.
  219. 7:31I’ll leave the link down in the
  220. 7:33description so you can discover Epidemic
  221. 7:35Sound, whom I also thank for sponsoring
  222. 7:37this video.
  223. 7:38Take advantage now because with Black
  224. 7:40Friday you will get a 50%discount on an
  225. 7:42annual plan with Epidemic Sound, getting
  226. 7:44the biggest discount of the year. And
  227. 7:46once we know this, let's continue with
  228. 7:47the next block.
  229. 7:52We are going to continue now by seeing
  230. 7:54how we can really get even more out of
  231. 7:56the new Gemini 4 Hydrogen model.
  232. 7:58Because although we have seen that we
  233. 8:00have three effort levels for Gemini, the
  234. 8:01truth is that we have an additional one
  235. 8:03that makes the difference.
  236. 8:04To discover all this, we have to jump to
  237. 8:07Google AI Studio, as this is the
  238. 8:08platform where we can use all the models
  239. 8:11in the Google ecosystem in a more
  240. 8:13unlocked way.
  241. 8:14In fact, notice that if we click on the
  242. 8:17top right, clicking on the latest models
  243. 8:19, we will find this latest one, Gemini 4
  244. 8:21Hydrogen, which we will be able to use,
  245. 8:23unlike its other two models, for free.
  246. 8:26I should warn you that this wouldn't be
  247. 8:28unlimited, just like with Gemini, but
  248. 8:30throughout the video we will see how we
  249. 8:33can use this model limitlessly.
  250. 8:35But going back to where we left off, I'm
  251. 8:37going to click on this model here and
  252. 8:38notice that for the reasoning level
  253. 8:40section, by default, we will have it in
  254. 8:42this case set to high, but we will also
  255. 8:44have a fourth level which is extra high
  256. 8:46.
  257. 8:47To put it to the test, I'm going to
  258. 8:48enter the exact same prompt we used for
  259. 8:50the octopus video in Gemini to see if
  260. 8:51there's any difference.
  261. 8:54To do this, I'll click on Run, and after
  262. 8:55about 3 minutes, we have all the HTML
  263. 8:57code right here.
  264. 8:59I'm going to click on the Preview
  265. 9:01section so we can open it, and we get a
  266. 9:03preview right here, but along with that
  267. 9:05, it shows we have Google voices
  268. 9:06connected, which should make it sound a
  269. 9:08bit more natural.
  270. 9:11A few more options also allow us to
  271. 9:13upload our own voice, but let's see if
  272. 9:14the video itself actually looks better.
  273. 9:17So, for that, I'm going to download this
  274. 9:20HTML code, open it to see it in a larger
  275. 9:22view, and see how it sounds and looks
  276. 9:25here.
  277. 9:25>> Did you know that this animal has three
  278. 9:27hearts?
  279. 9:29Two of them, the branchial hearts, pump
  280. 9:31blood to the gills to collect oxygen.
  281. 9:35And the third one pumps that oxygenated
  282. 9:36blood to the rest of the body.
  283. 9:40But when it swims, this last one stops
  284. 9:42beating. That's why swimming is so hard
  285. 9:44for them.
  286. 10:08>> settings, we will be able to use the
  287. 10:10Gemini 4 Hydrogen model, which we can
  288. 10:12use for free, and it will also offer us
  289. 10:15better results, since here we will have
  290. 10:17an even higher level of power.
  291. 10:20Notice that if we click on this section
  292. 10:22here, we'll have a max mode where we
  293. 10:24force this model to work for longer so
  294. 10:26that it offers us better results.
  295. 10:30To put this to the test directly, I'm
  296. 10:32going to do an example I've done in
  297. 10:33other models, which would be creating a
  298. 10:35game similar to GTA 6, that is, an
  299. 10:36open-world game, but with a somewhat
  300. 10:38simpler style.
  301. 10:41For this, in the prompt section, I would
  302. 10:43have entered this entire prompt here,
  303. 10:45where I'm telling it to create a
  304. 10:47playable 3D game inspired by open-world
  305. 10:49games, like a Minecraft or a GTA 6,
  306. 10:51where I want everything to appear:
  307. 10:53streets, cars, planes, so I can drive
  308. 10:56everything and it really lets me
  309. 10:57interact completely with this world.
  310. 11:01I'm going to click here on send so it
  311. 11:03starts working on this, and after about
  312. 11:05a little over 10 minutes, which is a
  313. 11:07long time for this model, let's see what
  314. 11:10it has managed to do for us.
  315. 11:12In fact, here I find this open map where
  316. 11:14I see a motorcycle appears, and from
  317. 11:17here I also see other people appearing,
  318. 11:19and the whole city. I also see different
  319. 11:22cars appearing here, for example a taxi
  320. 11:24, and if I click E, well, I can see that
  321. 11:26from here I can drive and it would all
  322. 11:28be working perfectly. Let's see if we
  323. 11:30can find, look, I'm just seeing a plane
  324. 11:31over here, let's see if we can interact
  325. 11:33with it.
  326. 11:34For that, I'm going to get out; it seems
  327. 11:36like all the buttons are working
  328. 11:37perfectly.
  329. 11:38I click E and from here, well, I could
  330. 11:40also be flying this plane.
  331. 11:43It is true that the movements might need
  332. 11:45a bit of iteration, but the truth is it
  333. 11:46seems to work quite well all things
  334. 11:48considered.
  335. 11:49Let's see if it also gave us anything
  336. 11:51else because I see something over there
  337. 11:52at the end. Okay, I was saying that I
  338. 11:55saw a port over here, so let's try to
  339. 11:57land this and from here we'll see if I
  340. 12:00can get on this boat, which I see is
  341. 12:03also working perfectly. Also, I can
  342. 12:05interact here.
  343. 12:07I also see a map where different little
  344. 12:09dots appear, which I understand are
  345. 12:11different missions, and ultimately, we
  346. 12:13have this simulation that has never had
  347. 12:15such a detailed result that works so
  348. 12:17well on the first try and in so little
  349. 12:19time, since usually these types of
  350. 12:21examples used to take around 30 or 40
  351. 12:24minutes.
  352. 12:25The Quicksol one took about 10 minutes
  353. 12:26and it did it much better than previous
  354. 12:28tests I was doing.
  355. 12:29In fact, notice that if we go back to
  356. 12:31the game, we could find other options
  357. 12:33like switching to night mode, where we
  358. 12:35could see everything perfectly applied.
  359. 12:37I also see different views like, for
  360. 12:39example, this first-person one where I'm
  361. 12:41going to try to get on a vehicle like
  362. 12:43this motorcycle, and look at how I could
  363. 12:45ride it from here.
  364. 12:47I even see many details like, for
  365. 12:49example, trucks or even cranes like this
  366. 12:51one here that I'm going to check by
  367. 12:52entering it and changing the view here
  368. 12:54to see if I can also operate this crane
  369. 12:56, and as you can see, I could also do
  370. 12:58that.
  371. 12:59And look at a result that was great, not
  372. 13:01only because the model is good, but
  373. 13:03because we also forced it to spend more
  374. 13:05time working in that max mode. In fact,
  375. 13:08while that part was being done, I tried
  376. 13:10to do it in Gemini too, and look here at
  377. 13:12how we have a much worse result where we
  378. 13:14really don't have any detail, no
  379. 13:15vehicles or people appear.
  380. 13:17I can kind of walk or levitate through
  381. 13:19this map here, but the truth is the
  382. 13:21result doesn't even come close to what
  383. 13:23we've done directly in Google AI Studio
  384. 13:26.
  385. 13:27But be careful, because this isn't just
  386. 13:29about us having used an exclusive
  387. 13:31application to create apps, since
  388. 13:32directly in Google AI Studio's chat mode
  389. 13:34I asked it for that same game, it
  390. 13:36created it in HTML, and if I open it
  391. 13:38here with that chat mode and a higher
  392. 13:40reasoning level, well, look here at how
  393. 13:42the result would actually be much better
  394. 13:44.
  395. 13:46I also find those people, I find cars
  396. 13:48that I will also be able to interact
  397. 13:50with right away, and all of this would
  398. 13:52be working perfectly, just like other
  399. 13:54means of transport such as, for example
  400. 13:56, that plane we are approaching right
  401. 13:58now.
  402. 14:00And if everything has worked this well
  403. 14:02and on the first try, well, from here
  404. 14:03I'll also be able to get on, and from
  405. 14:05here we could really see that it would
  406. 14:07also be working perfectly, since we
  407. 14:09could steer it as we can see here and we
  408. 14:11would be flying over this city.
  409. 14:17In this next part, we will see how we
  410. 14:19could actually extract this model from
  411. 14:21the Google ecosystem to be able to
  412. 14:22include it in any tool or harness we
  413. 14:24work with.
  414. 14:25Because yes, another thing they have
  415. 14:27released is the ability to use the
  416. 14:30Google Gemini API using this latest
  417. 14:32Gemini 4 Hydrogen model for free.
  418. 14:35To make this possible, I have returned
  419. 14:37to Google AI Studio, and from here, if
  420. 14:38we click on this key section to go to
  421. 14:40the API key, we will be able to find the
  422. 14:42API keys we have created or, conversely
  423. 14:44, if we didn't have any, by clicking on
  424. 14:46create API key, we will be able to
  425. 14:48generate one.
  426. 14:49But before taking this key and bringing
  427. 14:51the technology to another tool, notice
  428. 14:53that if we click on the rate limit
  429. 14:55section, we will be able to see from
  430. 14:57here the entire breakdown of models and
  431. 14:59how much we will be able to use them
  432. 15:00with their APIs for free.
  433. 15:03For example, here in the first one, we
  434. 15:05find Gemini 4 Hydrogen, from which they
  435. 15:07tell us we have an RPM, that is, a usage
  436. 15:09per minute of up to five requests.
  437. 15:12Then here we would have the input limit
  438. 15:14per minute in terms of tokens, which
  439. 15:15would be about 250,000, and how many
  440. 15:17requests we can make daily.
  441. 15:19In this case, we would have 20.
  442. 15:21This will actually change depending on
  443. 15:23the model, but keep in mind that this
  444. 15:25Gemini 4 has exactly the same limits as
  445. 15:27the small Gemini Flash models we had
  446. 15:29previously.
  447. 15:31Knowing this and returning to the API
  448. 15:33key, once we create an API key and click
  449. 15:35copy, we could now bring all this
  450. 15:37technology to any artificial
  451. 15:39intelligence tool, whether it's our own
  452. 15:41or, as in this case, I will be using a
  453. 15:43harness which will be DeepSeek's.
  454. 15:46Since this application here is
  455. 15:47open-source, it's quite similar to Open
  456. 15:50Claude, Hermes, or similar tools that
  457. 15:52allow us to connect any model we want to
  458. 15:54this technology. In fact, notice that if
  459. 15:57I click on the plus section, go into
  460. 15:59settings, and then head to the models
  461. 16:01section, if I click add a new provider
  462. 16:04from here and select the Google provider
  463. 16:06, by pasting the API key we just copied
  464. 16:08and clicking apply, it tells us that the
  465. 16:11models are now available. Notice that if
  466. 16:13I now click the model selector here and
  467. 16:16go to the Google provider, I can see all
  468. 16:18the models we have available.
  469. 16:20However, when I did this, by default,
  470. 16:23the Gemini 4 Hydrogen model didn't
  471. 16:25appear, because even though models like
  472. 16:27Gemini 4 Argon show up here, the truth
  473. 16:29is you have to pay to use them.
  474. 16:32If this happens to you too, there's a
  475. 16:34trick: if I go back to the settings
  476. 16:36section, return to models, and edit this
  477. 16:38provider with custom configuration,
  478. 16:40notice that when I click search for
  479. 16:42available models, that model we are
  480. 16:44talking about really doesn't appear.
  481. 16:48However, if I update it to show all
  482. 16:50these models here and add it manually by
  483. 16:52entering this ID, which I'll also leave
  484. 16:54below in the description—which is simply
  485. 16:56putting Gemini 4 Hydrogen—then it should
  486. 16:59work.
  487. 17:01From here, we would enter the name we
  488. 17:03want to appear when we select it; in
  489. 17:05this case, I'll put "Hydrogen Free," and
  490. 17:07if I click apply, I can now find this
  491. 17:09Google model I named "Hydrogen Free."
  492. 17:12And if I now say "Hello, can you hear me
  493. 17:14?"
  494. 17:15and click send, then about 2 seconds
  495. 17:17later it will tell us that it receives
  496. 17:20us correctly from the Deep Seek harness
  497. 17:22and that, effectively, it can work with
  498. 17:24us.
  499. 17:26With this, we can really take this free
  500. 17:28technology, which has daily limits, and
  501. 17:30use it for free in other places beyond
  502. 17:33the Google ecosystem, having a very
  503. 17:35powerful model that is also very fast.
  504. 17:38But, at the beginning of the video,
  505. 17:39besides saying it could be used for free
  506. 17:41, I also told you it was unlimited. To
  507. 17:43see this, let's move on to the next
  508. 17:44block.
  509. 17:48In this penultimate block, we'll see how
  510. 17:50we can manage to use this latest Gemini
  511. 17:524 model for free, but also in an
  512. 17:54unlimited way.
  513. 17:55There is something here that hasn’t even
  514. 17:58been officially stated, but Google's AI
  515. 18:00mode is actually already using this
  516. 18:02latest model that Gemini has released,
  517. 18:05only here we don't have any kind of
  518. 18:07limit.
  519. 18:08For this, I’ve gone back to Google, but
  520. 18:10if I now click on AI mode from here and
  521. 18:12in the prompt section I tell it that I
  522. 18:14want to visually understand how a home
  523. 18:16in Spain consumes energy with a bit more
  524. 18:18information, when I click send, it will
  525. 18:20start processing this, and in just a few
  526. 18:22seconds, not only will it be giving me
  527. 18:24this information, but I’ll also have
  528. 18:26this interactive visual to understand it
  529. 18:28much better.
  530. 18:30Notice how here I can turn on, for
  531. 18:32example, the heating and the dryer, so
  532. 18:34the meter would be increasing, and as I
  533. 18:36deactivate things, this meter would also
  534. 18:38go down.
  535. 18:39Additionally, I can tell it here how
  536. 18:41much on average each of these things
  537. 18:43would be using so that when I click turn
  538. 18:45on all, it shows me what the meter would
  539. 18:47read, what the estimated distribution of
  540. 18:49monthly expenses would be, and in short
  541. 18:51, the whole application to understand
  542. 18:53this easily.
  543. 18:55And besides that, we get much faster
  544. 18:57answers here thanks to having this
  545. 18:58latest Gemini 4 model.
  546. 19:01In fact, to verify this, I am going to
  547. 19:04click to start a new chat within
  548. 19:06Google's AI, and if I ask it here if it
  549. 19:09could confirm which model it is using,
  550. 19:11I’ll hit send, and it would tell me that
  551. 19:14it is indeed using Gemini 4 Hydrogen
  552. 19:16directly in this chat, which, unlike
  553. 19:19Gemini, doesn't have the limits we find
  554. 19:21in that section where we would have 5-
  555. 19:24hour or weekly limits, since inside
  556. 19:26Google's AI mode, at least publicly,
  557. 19:30there are no limits at all.
  558. 19:32And from here, the whole process of
  559. 19:34searching across the entire internet is
  560. 19:35also optimized.
  561. 19:37In fact, let’s do another use case, and
  562. 19:39for this, from a new chat, I’m going to
  563. 19:40enter the following prompt where I’ll
  564. 19:42ask it to compare what it would be like
  565. 19:44to live in Madrid, Barcelona, Valencia,
  566. 19:46Málaga, and Jaén for someone who works
  567. 19:47from home so that I can see what
  568. 19:49purchasing power I would need to live
  569. 19:51comfortably.
  570. 19:53I click send, and it will start
  571. 19:55processing so that, shortly after, not
  572. 19:57only will it give us an answer based on
  573. 19:59real internet information, but beyond
  574. 20:02all this data and interactive maps, we
  575. 20:04will be able to find an interactive
  576. 20:06panel here where, even if the map of
  577. 20:08Spain isn't correct, it would tell us
  578. 20:10roughly where each location is and the
  579. 20:13approximate budget we would need.
  580. 20:17For instance, if I change it here to say
  581. 20:18I'd like to live in a family apartment
  582. 20:20where I do need to use a vehicle daily,
  583. 20:22it would tell me that with those monthly
  584. 20:23euros, I couldn't live anywhere except
  585. 20:25Jaén.
  586. 20:26And, honestly, it would be a bit tight.
  587. 20:29If I now increase the budget, look at
  588. 20:31how other locations start turning green
  589. 20:33and the green bar appears a bit low,
  590. 20:34meaning I'm more comfortable in those
  591. 20:36locations.
  592. 20:38And this way, we can easily see all that
  593. 20:40kind of information that would take us
  594. 20:42quite a long time to research.
  595. 20:45Mind you, we also have to talk about the
  596. 20:47downsides, because even though it's
  597. 20:48completely unlimited, these models are
  598. 20:50the smallest and fastest ones for
  599. 20:51searching and finding information as
  600. 20:53quickly as possible.
  601. 20:54So, in Google's AI mode, we wouldn't be
  602. 20:56using the most advanced reasoning that
  603. 20:58provides better results for complex
  604. 21:00tasks.
  605. 21:01In fact, to see what this translates to
  606. 21:03, from a new chat, I asked it to create
  607. 21:05that interactive open-world simulator
  608. 21:07game we saw earlier. And while it's true
  609. 21:09that it did it in record time, look at
  610. 21:11how the results really don't come close
  611. 21:14to what we saw previously. For example,
  612. 21:16if I click here on "change vehicle,"
  613. 21:18this car would indeed be working.
  614. 21:20As we can see here, if I click on "
  615. 21:22change vehicle," the plane would be
  616. 21:23flying, but it's something very simple
  617. 21:25that isn't really comparable to the
  618. 21:27application we saw before in Google AI
  619. 21:29Studio.
  620. 21:33Let's finish with a bonus.
  621. 21:35Just a few weeks ago, Google launched a
  622. 21:37new platform that lets us create video
  623. 21:39games for free, and it has just been
  624. 21:41updated with the latest Gemini 1.5 Pro
  625. 21:43model.
  626. 21:45To see this, we would have to go to
  627. 21:47Playground, which is this platform here
  628. 21:49that we can access for free, but there
  629. 21:51is a slight catch. And that is the fact
  630. 21:53that it can only be used in the United
  631. 21:55States. In my case, I'm using it from
  632. 21:57Spain because I have a VPN that is
  633. 21:59currently simulating that I'm in New
  634. 22:01York.
  635. 22:02Regarding the VPN, keep in mind that
  636. 22:04even though they aren't sponsoring this
  637. 22:05video, I'll still leave the one I use in
  638. 22:07the description below, since with Black
  639. 22:08Friday now, there are some great prices
  640. 22:10, especially with a coupon I'll share
  641. 22:11below.
  642. 22:13Knowing this, look at how a bunch of
  643. 22:15games appear right here, like Street
  644. 22:18scale, adventure, car, Monopoly,
  645. 22:20Minecraft-style, and zombie games.
  646. 22:23Or one that really surprised me was this
  647. 22:25motocross one, where if I click on "
  648. 22:27play now," we'll see a game that someone
  649. 22:30made using artificial intelligence. Here
  650. 22:32, to see this, I'm going to click play,
  651. 22:34we'll hit freestyle, I'll go to
  652. 22:36customize garage, and look at how this
  653. 22:39bike appears here as if it were a
  654. 22:41completely full game where I can change
  655. 22:43the bike's color. I'm going to leave it
  656. 22:45in red. I can also change the rider and
  657. 22:47which engine I want to use.
  658. 22:49In this case, I'll use this four-stroke
  659. 22:50one, along with the tricks I can do and
  660. 22:52what the controls would be.
  661. 22:55And if I click on start session, look at
  662. 22:57how we'd be with this bike, which is
  663. 22:58also totally playable.
  664. 23:00In fact, let's try to do a trick.
  665. 23:02Look at how it's working perfectly here
  666. 23:04.
  667. 23:05And it really is surprising that this
  668. 23:07was coded by just anyone. Actually,
  669. 23:09doing it is very simple. For that, if I
  670. 23:11click back here and click on the create
  671. 23:13section, from here we can specify any
  672. 23:16game we want to create. To do that,
  673. 23:18continuing with the motorcycle example,
  674. 23:20I thought of creating a game I used to
  675. 23:21be very addicted to, which was the
  676. 23:23typical 2D motocross game that had
  677. 23:24different obstacles you had to pass. To
  678. 23:26do this, I would just enter that prompt
  679. 23:29with all these requirements. I'm going
  680. 23:31to tell it I want it in 2D graphics, for
  681. 23:32one person, although notice that we
  682. 23:34could also make multiplayer games.
  683. 23:36I click send, and after a few minutes
  684. 23:38and it having created all the things it
  685. 23:39was asking me about—I told it to
  686. 23:41generate everything for me—I'd already
  687. 23:42have this video game here that look at
  688. 23:44how I'm going to be able to play.
  689. 23:46And from here, we have a visualization
  690. 23:48where we can even change the different
  691. 23:50levels, and from here, we can see it
  692. 23:51would really be very similar to these
  693. 23:53types of games.
  694. 23:55This is a platform that could really
  695. 23:56support another video, so if you are
  696. 23:58interested in learning more about it in
  697. 23:59detail, let me know in the comments so I
  698. 24:01can make one about it. With this, Google
  699. 24:03seems to have once again democratized
  700. 24:05the use of artificial intelligence with
  701. 24:07its latest accessible model. And this
  702. 24:09shows us that, in many cases, to move
  703. 24:10forward, sometimes you have to go back
  704. 24:12to the beginning.

About this transcript

This page contains the full transcript of Google ha Vuelto! Lanzó GEMINI 4 Hydrogen al fin GRATIS y SIN Límites by Alejavi Rivera, generated from the public captions YouTube serves with the video. The transcript has 4,757 words across 704 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.