1 00:00:00,240 --> 00:00:02,480 So, looks like we just got flashed by 2 00:00:02,480 --> 00:00:04,640 Google DeepMind [music] once again 3 00:00:04,640 --> 00:00:06,879 because they just dropped Gemini 3.7 4 00:00:06,879 --> 00:00:09,440 Flash today. Yes, their newest model, 5 00:00:09,440 --> 00:00:11,679 which is once again a Flash model, is 6 00:00:11,679 --> 00:00:14,160 here. And this is only after 3 weeks 7 00:00:14,160 --> 00:00:17,279 since Gemini 3.6 Flash. And what they 8 00:00:17,279 --> 00:00:20,000 are saying is that Gemini 3.7 Flash is 9 00:00:20,000 --> 00:00:22,800 their most intelligent workhorse model. 10 00:00:22,800 --> 00:00:25,359 And this is coming 3 weeks after 3.6 six 11 00:00:25,359 --> 00:00:27,840 flash because of developer feedback and 12 00:00:27,840 --> 00:00:30,480 algorithmic innovations. And that is why 13 00:00:30,480 --> 00:00:32,480 they're saying the Gemini 3.7 Flash 14 00:00:32,480 --> 00:00:34,320 model is, you know, out with so much 15 00:00:34,320 --> 00:00:36,160 improvement cuz if we take a look at the 16 00:00:36,160 --> 00:00:38,399 benchmarks, this model is way better 17 00:00:38,399 --> 00:00:41,360 than the Gemini 3.6 Flash. But I think 18 00:00:41,360 --> 00:00:43,440 this is also because we know internally 19 00:00:43,440 --> 00:00:44,879 that they're planning on cancelelling 20 00:00:44,879 --> 00:00:48,160 Gemini 3.5 Pro completely and working on 21 00:00:48,160 --> 00:00:50,480 Gemini 4 at the moment. So maybe the 22 00:00:50,480 --> 00:00:51,920 improvements they made with the Pro 23 00:00:51,920 --> 00:00:53,520 model that they're not going to be 24 00:00:53,520 --> 00:00:55,840 dropping anymore. Maybe they repackaged 25 00:00:55,840 --> 00:00:58,320 it as a flash model because this model 26 00:00:58,320 --> 00:01:01,199 is better than 3.6 and if the timeline 27 00:01:01,199 --> 00:01:03,120 is three weeks plus they're planning on 28 00:01:03,120 --> 00:01:05,519 cancelling the 3.5 Pro, it's quite 29 00:01:05,519 --> 00:01:07,600 likely that is the case. We'll just 30 00:01:07,600 --> 00:01:09,119 ignore that and put that into the side. 31 00:01:09,119 --> 00:01:10,479 We don't know when we're getting a new 32 00:01:10,479 --> 00:01:12,080 Pro model from the Google Deep Mind 33 00:01:12,080 --> 00:01:14,400 team, but this model across many of the 34 00:01:14,400 --> 00:01:17,119 benchmarks we will see that yes, it is 35 00:01:17,119 --> 00:01:19,920 stronger than the 3.6 Flash model. For 36 00:01:19,920 --> 00:01:21,840 example, when it comes to code quality, 37 00:01:21,840 --> 00:01:24,159 production code quality, the model is 38 00:01:24,159 --> 00:01:26,159 producing 43.6% 39 00:01:26,159 --> 00:01:29,680 versus the flash model 34.4%. 40 00:01:29,680 --> 00:01:33,520 And then Sonet 5 is at 42.7% 41 00:01:33,520 --> 00:01:36,799 and the Terra model is at 41.3%. 42 00:01:36,799 --> 00:01:38,799 So, one thing to remember, since this is 43 00:01:38,799 --> 00:01:40,240 a flash model, you're not going to see 44 00:01:40,240 --> 00:01:42,159 them compare this to Opus 5 or any of 45 00:01:42,159 --> 00:01:44,240 the other stronger models. The reason 46 00:01:44,240 --> 00:01:46,240 being cuz this is not their strongest 47 00:01:46,240 --> 00:01:49,119 tier. and the 3.1 Pro model when they 48 00:01:49,119 --> 00:01:51,600 finally decide to upgrade it to Gemini 4 49 00:01:51,600 --> 00:01:54,479 Pro, then we'll see it being compared to 50 00:01:54,479 --> 00:01:57,920 Opus 5 or GPT 5.6 Soul. But anyways, we 51 00:01:57,920 --> 00:01:59,280 can see that this model is an 52 00:01:59,280 --> 00:02:01,520 improvement from 3.6 Flash and I would 53 00:02:01,520 --> 00:02:04,560 say that's basically the biggest result 54 00:02:04,560 --> 00:02:06,479 or the biggest update that we saw with 55 00:02:06,479 --> 00:02:08,560 the new model because it does not really 56 00:02:08,560 --> 00:02:10,720 like change up things a lot is still 57 00:02:10,720 --> 00:02:12,560 like not becoming the number one model 58 00:02:12,560 --> 00:02:15,040 or the number one flash model. Even on 59 00:02:15,040 --> 00:02:17,200 some benchmarks, Deepseek version 4 60 00:02:17,200 --> 00:02:19,920 flash is actually cheaper and more 61 00:02:19,920 --> 00:02:22,160 intelligent than this model. But let's 62 00:02:22,160 --> 00:02:23,680 just take a look at the benchmarks that 63 00:02:23,680 --> 00:02:25,840 they have published. On Long Horizon 64 00:02:25,840 --> 00:02:28,080 software engineering, this model is a 65 00:02:28,080 --> 00:02:30,959 little bit behind GPT 5.6 Terra, which 66 00:02:30,959 --> 00:02:32,959 sits at 69.6% 67 00:02:32,959 --> 00:02:36,400 and then 65.3% is the Flash model. The 68 00:02:36,400 --> 00:02:40,800 older Flash model 3.6 sits at 48.6%. 69 00:02:40,800 --> 00:02:44,000 Sonic 5 sits at 53.8% 8%. And the new 70 00:02:44,000 --> 00:02:46,160 player that is finally being included on 71 00:02:46,160 --> 00:02:48,000 benchmarks, which even the Google 72 00:02:48,000 --> 00:02:49,920 DeepMind team is considering with their 73 00:02:49,920 --> 00:02:52,480 launch, is Muse Spark 1.2, which is 74 00:02:52,480 --> 00:02:54,640 sitting at 54.9%. 75 00:02:54,640 --> 00:02:57,519 So, welcome Meta to the benchmark charts 76 00:02:57,519 --> 00:02:59,360 because now we're starting to see it 77 00:02:59,360 --> 00:03:01,519 appear on more and more benchmarks. And 78 00:03:01,519 --> 00:03:02,800 if we also take a look at web 79 00:03:02,800 --> 00:03:05,360 development, this model is getting a ELO 80 00:03:05,360 --> 00:03:08,159 score of 1588 versus their old model is 81 00:03:08,159 --> 00:03:10,319 1538. So, not a crazy difference, but 82 00:03:10,319 --> 00:03:11,840 compared to everything else out there, 83 00:03:11,840 --> 00:03:13,360 this is number one. When I say 84 00:03:13,360 --> 00:03:14,879 everything else out there, once again, 85 00:03:14,879 --> 00:03:16,720 compared to all the mid-tier models, 86 00:03:16,720 --> 00:03:18,640 this is beating all of them when it 87 00:03:18,640 --> 00:03:20,640 comes to web development. Now, this 88 00:03:20,640 --> 00:03:22,480 model is probably going to be used in 89 00:03:22,480 --> 00:03:24,159 enterprises a lot just because of 90 00:03:24,159 --> 00:03:26,000 Google's footprint in the enterprise 91 00:03:26,000 --> 00:03:28,159 space, but we are seeing this model 92 00:03:28,159 --> 00:03:31,599 achieve on the automation bench 30.4%. 93 00:03:31,599 --> 00:03:35,440 And GPT 5.6 Terra sits at 23.6%. 94 00:03:35,440 --> 00:03:37,599 So, yes, this model is stronger than the 95 00:03:37,599 --> 00:03:39,760 other Flash models out there, but as I 96 00:03:39,760 --> 00:03:41,440 said, they haven't included Deepseek 97 00:03:41,440 --> 00:03:43,599 version for Flash because if they do, in 98 00:03:43,599 --> 00:03:45,519 some areas, that model is actually quite 99 00:03:45,519 --> 00:03:48,480 better than the 3.7 Flash model. Before 100 00:03:48,480 --> 00:03:50,319 we continue, if you're building AI 101 00:03:50,319 --> 00:03:52,480 agents or just messing around with them, 102 00:03:52,480 --> 00:03:54,799 Arcade is worth knowing about. It's the 103 00:03:54,799 --> 00:03:56,799 runtime that lets your agent actually do 104 00:03:56,799 --> 00:03:58,319 things instead of just talking about 105 00:03:58,319 --> 00:04:00,560 them because that's the gap right now. 106 00:04:00,560 --> 00:04:02,799 The models are smart enough. Your agent 107 00:04:02,799 --> 00:04:04,720 can figure out exactly what needs to 108 00:04:04,720 --> 00:04:06,560 happen in your email, your Slack, your 109 00:04:06,560 --> 00:04:09,680 CRM. It just can't go in and do it. And 110 00:04:09,680 --> 00:04:11,519 the reason isn't intelligence, it's 111 00:04:11,519 --> 00:04:13,680 permissions. Something has to prove the 112 00:04:13,680 --> 00:04:15,840 agent is allowed to act on behalf of a 113 00:04:15,840 --> 00:04:18,479 specific person in a specific account. 114 00:04:18,479 --> 00:04:20,239 That's the messy part everyone runs 115 00:04:20,239 --> 00:04:22,479 into, and it's the partit actually 116 00:04:22,479 --> 00:04:24,320 handles for you. So instead of your 117 00:04:24,320 --> 00:04:26,240 agent using one shared login for 118 00:04:26,240 --> 00:04:28,160 everybody, it acts as whoever is 119 00:04:28,160 --> 00:04:30,320 actually signed in with exactly the 120 00:04:30,320 --> 00:04:32,560 access that person has. If they can't 121 00:04:32,560 --> 00:04:34,880 see something, the agent can't either. 122 00:04:34,880 --> 00:04:36,479 And you never have to touch any of that 123 00:04:36,479 --> 00:04:38,960 setup yourself. Then there's the tools. 124 00:04:38,960 --> 00:04:40,800 Arcade has thousands of them already 125 00:04:40,800 --> 00:04:43,120 built for Gmail, Google Drive, Slack, 126 00:04:43,120 --> 00:04:45,520 Notion, Salesforce, most of the apps 127 00:04:45,520 --> 00:04:47,360 people already work in, and they are 128 00:04:47,360 --> 00:04:49,759 built specifically for AI to use. So the 129 00:04:49,759 --> 00:04:51,440 agent gets it right the first time 130 00:04:51,440 --> 00:04:53,040 instead of guessing and failing and 131 00:04:53,040 --> 00:04:55,120 retrying. It also keeps a record of 132 00:04:55,120 --> 00:04:57,520 everything, what the agent did, for who 133 00:04:57,520 --> 00:04:59,440 and where, which matters a lot the 134 00:04:59,440 --> 00:05:00,880 moment other people start using the 135 00:05:00,880 --> 00:05:02,639 thing you built. And it works with 136 00:05:02,639 --> 00:05:04,560 whatever you're already using. Any 137 00:05:04,560 --> 00:05:06,960 model, any framework, cloud, cursor, 138 00:05:06,960 --> 00:05:09,520 chat, GPT, doesn't matter. So you're not 139 00:05:09,520 --> 00:05:11,139 just giving an AI a list of tools and 140 00:05:11,139 --> 00:05:12,720 [music] hoping it works. You're giving 141 00:05:12,720 --> 00:05:14,800 it a place where it can safely take real 142 00:05:14,800 --> 00:05:17,360 actions in real apps. It's free to start 143 00:05:17,360 --> 00:05:19,120 and the link is in the description. 144 00:05:19,120 --> 00:05:20,880 Thank you once again for Arcade for 145 00:05:20,880 --> 00:05:22,960 sponsoring today's video. Now, let's get 146 00:05:22,960 --> 00:05:25,440 back into the video. Now, one thing to 147 00:05:25,440 --> 00:05:28,560 note is that the Gemini 3.7 Flash model 148 00:05:28,560 --> 00:05:30,320 through the end of this year, so end of 149 00:05:30,320 --> 00:05:33,360 2026, they have a cheap pricing model 150 00:05:33,360 --> 00:05:36,160 that they're placing on the model. 75 151 00:05:36,160 --> 00:05:39,840 per 1 million input tokens and $3.75 for 152 00:05:39,840 --> 00:05:41,680 1 million output tokens. So, it's a 153 00:05:41,680 --> 00:05:43,759 competitive price, but this is only for 154 00:05:43,759 --> 00:05:45,919 the next 6 months. Because after those 155 00:05:45,919 --> 00:05:48,240 six months are done, the model's pricing 156 00:05:48,240 --> 00:05:50,160 is actually, you know, a little bit more 157 00:05:50,160 --> 00:05:51,919 expensive. And now they show it at the 158 00:05:51,919 --> 00:05:54,720 bottom over here, you can see that after 159 00:05:54,720 --> 00:05:58,160 starting January of 2027, it will become 160 00:05:58,160 --> 00:06:02,479 $1.50 per input and $7.50 per output. So 161 00:06:02,479 --> 00:06:04,720 yeah, it's still cheap compared to the 162 00:06:04,720 --> 00:06:06,560 other frontier labs, but it's not as 163 00:06:06,560 --> 00:06:08,639 cheap as, for example, MU Spark when the 164 00:06:08,639 --> 00:06:11,199 pricing is updated or even the Deepsee 165 00:06:11,199 --> 00:06:13,280 version for Flash. But across these 166 00:06:13,280 --> 00:06:15,440 benchmarks, we can see that this model 167 00:06:15,440 --> 00:06:17,360 is better in many areas that they 168 00:06:17,360 --> 00:06:18,880 highlighted at the top. But then they 169 00:06:18,880 --> 00:06:21,199 also have some other areas like long 170 00:06:21,199 --> 00:06:22,960 video understanding, which this model 171 00:06:22,960 --> 00:06:27,199 excels at 85.4% versus 78.9% 172 00:06:27,199 --> 00:06:29,120 for the Terra model. And then the old 173 00:06:29,120 --> 00:06:30,880 model was also pretty good at that, 174 00:06:30,880 --> 00:06:32,550 84.2%. 175 00:06:32,550 --> 00:06:32,560 84.2%. 176 00:06:32,560 --> 00:06:34,800 Then long context performance the model 177 00:06:34,800 --> 00:06:39,440 is at 97% and this model the GPT 6 Terra 178 00:06:39,440 --> 00:06:42,160 one it sits at 93.5%. 179 00:06:42,160 --> 00:06:44,160 So yeah this model in summary it is 180 00:06:44,160 --> 00:06:46,479 better so it's not all negative but it's 181 00:06:46,479 --> 00:06:48,639 not all like you know that positive 182 00:06:48,639 --> 00:06:50,479 where you are super excited for Google 183 00:06:50,479 --> 00:06:52,240 Deep Mind because as I said they're 184 00:06:52,240 --> 00:06:53,840 probably still holding off their biggest 185 00:06:53,840 --> 00:06:56,319 release for Gemini 4 lineup. Now, one 186 00:06:56,319 --> 00:06:58,160 thing people might have missed in their 187 00:06:58,160 --> 00:06:59,759 charts because these charts sometimes 188 00:06:59,759 --> 00:07:02,800 are so messy to read and understand, but 189 00:07:02,800 --> 00:07:06,160 the 3.7 flash is worse than GPT 5.6 190 00:07:06,160 --> 00:07:08,160 Luna, which is the model over here, 191 00:07:08,160 --> 00:07:10,560 which is achieving a higher score on 192 00:07:10,560 --> 00:07:13,199 this benchmark deepware engineering for 193 00:07:13,199 --> 00:07:15,520 about three times the cost. So, yeah, 194 00:07:15,520 --> 00:07:17,039 this is kind of interesting because 195 00:07:17,039 --> 00:07:19,599 yeah, the cost for Gemini 3.7 Flash is a 196 00:07:19,599 --> 00:07:21,440 little bit more than what it looks like. 197 00:07:21,440 --> 00:07:23,360 Now, some people are a little bit upset 198 00:07:23,360 --> 00:07:25,120 and they're like, "Oh, disgraceful. 199 00:07:25,120 --> 00:07:27,520 Google left out soul, opus, and fable 200 00:07:27,520 --> 00:07:29,680 because Google is incredibly behind. But 201 00:07:29,680 --> 00:07:31,039 I think one thing we got to remember, 202 00:07:31,039 --> 00:07:33,120 guys, is that this model is a flash 203 00:07:33,120 --> 00:07:35,919 model. It's not trying to be a pro model 204 00:07:35,919 --> 00:07:38,000 or it's not trying to compete with Opus 205 00:07:38,000 --> 00:07:40,639 or Soul or Fable 5 level models yet. 206 00:07:40,639 --> 00:07:42,800 That is probably going to be Gemini 4. 207 00:07:42,800 --> 00:07:44,880 So when Gemini 4 comes out, then I think 208 00:07:44,880 --> 00:07:47,199 it's okay for us to criticize them if 209 00:07:47,199 --> 00:07:49,440 they don't include Soul, Opus, and Fable 210 00:07:49,440 --> 00:07:51,199 in their benchmark charts because for 211 00:07:51,199 --> 00:07:53,199 now, I think what they have done is 212 00:07:53,199 --> 00:07:55,120 pretty accurate. One lab that I would 213 00:07:55,120 --> 00:07:56,639 have liked to see or one model for 214 00:07:56,639 --> 00:07:58,000 example I would have liked to see on 215 00:07:58,000 --> 00:07:59,840 that chart would be Deepseek version 4 216 00:07:59,840 --> 00:08:01,520 Flash because that would kind of spoil 217 00:08:01,520 --> 00:08:03,520 their release because that model is way 218 00:08:03,520 --> 00:08:06,319 cheaper compared to the Gemini 3.7 flash 219 00:08:06,319 --> 00:08:08,800 model lineup. Now this model jumped from 220 00:08:08,800 --> 00:08:11,199 number 19 to 8 on the web development 221 00:08:11,199 --> 00:08:13,599 area and we see it over here now and 222 00:08:13,599 --> 00:08:15,120 couple of models that are ahead of it 223 00:08:15,120 --> 00:08:18,160 are Opus 5 obviously Kim K3 Quinn 3.8 224 00:08:18,160 --> 00:08:21,680 Max Cloud Opus 5 Gro 4.6 6, which is a 225 00:08:21,680 --> 00:08:23,360 model that came yesterday, which was a 226 00:08:23,360 --> 00:08:26,720 big win for SpaceX, Fable 5, and 5.6. 227 00:08:26,720 --> 00:08:28,960 Soul. So, yeah, this model is trying to 228 00:08:28,960 --> 00:08:31,120 compete in the web development, but it's 229 00:08:31,120 --> 00:08:33,120 still behind all of these models, which 230 00:08:33,120 --> 00:08:35,120 is, you know, expected cuz it's a flash 231 00:08:35,120 --> 00:08:37,279 model. It's not really a pro model. 232 00:08:37,279 --> 00:08:39,599 Today, OpenAI has also launched a weight 233 00:08:39,599 --> 00:08:42,479 list for 5.6 so ultra fast mode. Now, 234 00:08:42,479 --> 00:08:44,959 this is possible because of the fast 235 00:08:44,959 --> 00:08:46,640 chips that they have access to now 236 00:08:46,640 --> 00:08:49,839 through Cabus and GPT 5.6 6o with this 237 00:08:49,839 --> 00:08:52,160 chip kind of running it is able to 238 00:08:52,160 --> 00:08:54,000 achieve an ultra fast mode that 239 00:08:54,000 --> 00:08:57,120 generates up to 750 output tokens per 240 00:08:57,120 --> 00:09:00,399 second which is about 14 times faster 241 00:09:00,399 --> 00:09:02,399 than the standard mode. So we're getting 242 00:09:02,399 --> 00:09:06,320 a really fast version of GPT 5.6 so now 243 00:09:06,320 --> 00:09:08,880 this is supposed to be used for live or 244 00:09:08,880 --> 00:09:10,720 near production workloads like you know 245 00:09:10,720 --> 00:09:13,279 real-time voice support commerce what 246 00:09:13,279 --> 00:09:15,839 this allows GPT 5.6 six soul to do is 247 00:09:15,839 --> 00:09:17,680 kind of be really fast in critical 248 00:09:17,680 --> 00:09:19,519 situations when people might be 249 00:09:19,519 --> 00:09:21,680 interacting with the AI agent like 250 00:09:21,680 --> 00:09:24,640 financial research security response 251 00:09:24,640 --> 00:09:26,480 support I think is going to be a big 252 00:09:26,480 --> 00:09:29,360 area where this new ultraast mode will 253 00:09:29,360 --> 00:09:32,000 be kind of implemented business and 254 00:09:32,000 --> 00:09:33,920 developer agents maybe yeah but I think 255 00:09:33,920 --> 00:09:35,600 like support or near production 256 00:09:35,600 --> 00:09:37,760 workloads like real-time voice I see 257 00:09:37,760 --> 00:09:40,160 this model really excelling at that and 258 00:09:40,160 --> 00:09:42,160 obviously this is still a weightless 259 00:09:42,160 --> 00:09:44,160 mode we don't know how many people are 260 00:09:44,160 --> 00:09:45,760 going to get access to this, how 261 00:09:45,760 --> 00:09:47,920 successful it is or what the pricing is. 262 00:09:47,920 --> 00:09:49,440 I don't know if the pricing has changed 263 00:09:49,440 --> 00:09:51,200 because there's no information in their 264 00:09:51,200 --> 00:09:54,000 actual, you know, blog post. But as I 265 00:09:54,000 --> 00:09:55,920 mentioned, couple of areas where they 266 00:09:55,920 --> 00:09:57,360 mentioned that this is going to be 267 00:09:57,360 --> 00:09:59,360 really important. Customer support and 268 00:09:59,360 --> 00:10:01,200 voice, commerce, live research and 269 00:10:01,200 --> 00:10:03,760 experimentation, financial research and 270 00:10:03,760 --> 00:10:05,839 security, incident response and 271 00:10:05,839 --> 00:10:07,760 reliability. But yeah, I think this 272 00:10:07,760 --> 00:10:09,680 partnership is going to be important for 273 00:10:09,680 --> 00:10:12,240 OpenAI going forward. But as I said, 14 274 00:10:12,240 --> 00:10:14,240 times the speed. What does that mean for 275 00:10:14,240 --> 00:10:16,800 cost? We don't know yet. But that's it 276 00:10:16,800 --> 00:10:18,560 for today's video. Make sure you guys 277 00:10:18,560 --> 00:10:20,480 are subscribed to the channel. Follow 278 00:10:20,480 --> 00:10:22,160 our new newsletter as well at 279 00:10:22,160 --> 00:10:24,389 universeai.behive.com 280 00:10:24,389 --> 00:10:24,399 universeai.behive.com 281 00:10:24,399 --> 00:10:26,079 as well as subscribe to the main channel 282 00:10:26,079 --> 00:10:28,560 World of AI and support us on X by 283 00:10:28,560 --> 00:10:30,800 following the Universe of AIZ as well. 284 00:10:30,800 --> 00:10:32,399 Until then, I'll see you guys in the 285 00:10:32,399 --> 00:10:34,560 next