1
00:00:00,080 --> 00:00:02,639
So, Deepseek version 4 Pro is officially

2
00:00:02,639 --> 00:00:04,799
out today. Now, you might be confused

3
00:00:04,799 --> 00:00:06,560
because this model was technically out,

4
00:00:06,560 --> 00:00:08,800
but it was in general availability, but

5
00:00:08,800 --> 00:00:10,400
this is the official release, meaning

6
00:00:10,400 --> 00:00:12,240
that they have trained the model a

7
00:00:12,240 --> 00:00:14,240
little bit more and produced a stronger

8
00:00:14,240 --> 00:00:16,000
version of the model that is available

9
00:00:16,000 --> 00:00:17,520
today. Now, if I were to summarize

10
00:00:17,520 --> 00:00:19,520
today's release in one simple sentence,

11
00:00:19,520 --> 00:00:21,520
it is that the price to performance is

12
00:00:21,520 --> 00:00:23,760
becoming a very, very important thing

13
00:00:23,760 --> 00:00:25,439
because not only do we have a new model

14
00:00:25,439 --> 00:00:27,680
from the Deepseek team, the SpaceX team

15
00:00:27,680 --> 00:00:30,240
also dropped Grock 4.6 six. And both of

16
00:00:30,240 --> 00:00:32,000
these models are competing with the best

17
00:00:32,000 --> 00:00:34,079
Frontier Labs at a fraction of their

18
00:00:34,079 --> 00:00:37,040
cost. Let's start with Gro 4.6. And one

19
00:00:37,040 --> 00:00:38,800
thing I'm going to say is that I'm

20
00:00:38,800 --> 00:00:40,879
genuinely surprised by the SpaceX team.

21
00:00:40,879 --> 00:00:42,480
I'm not trying to glaze them or Elon

22
00:00:42,480 --> 00:00:44,079
Musk or anything. I was just very

23
00:00:44,079 --> 00:00:46,480
critical of this lab. In 2025, they

24
00:00:46,480 --> 00:00:48,320
dropped Grog 4 and other models like

25
00:00:48,320 --> 00:00:50,399
that, but I wasn't really, you know,

26
00:00:50,399 --> 00:00:52,239
mind blown with their performance

27
00:00:52,239 --> 00:00:53,840
because they're pretty they're pretty

28
00:00:53,840 --> 00:00:56,000
subpar compared to any of the other labs

29
00:00:56,000 --> 00:00:58,480
out there. But in 2026, it looks like

30
00:00:58,480 --> 00:01:00,079
things have kind of changed a little

31
00:01:00,079 --> 00:01:02,399
because Grock 4.5 was quite competitive

32
00:01:02,399 --> 00:01:05,600
based off what it cost and Grock 4.6 is

33
00:01:05,600 --> 00:01:07,920
actually not too bad. If we take a look

34
00:01:07,920 --> 00:01:10,000
at the benchmarks, for example, if we

35
00:01:10,000 --> 00:01:11,920
start with the artificial analysis

36
00:01:11,920 --> 00:01:14,240
intelligence index, this model achieves

37
00:01:14,240 --> 00:01:18,159
a 61 and Fable 5 is at 62. Now, what's

38
00:01:18,159 --> 00:01:19,680
really important to remember is that

39
00:01:19,680 --> 00:01:22,159
once again, this model is quite cheap

40
00:01:22,159 --> 00:01:24,400
compared to Fable 5. This is about $2

41
00:01:24,400 --> 00:01:26,960
per million input tokens and $6 per

42
00:01:26,960 --> 00:01:29,439
million output tokens. In Fable 5 sits

43
00:01:29,439 --> 00:01:32,240
at $10 per million input tokens and $50

44
00:01:32,240 --> 00:01:34,400
per million output tokens. And this is

45
00:01:34,400 --> 00:01:36,400
why you start to appreciate this release

46
00:01:36,400 --> 00:01:38,320
a little bit more. It might not beat the

47
00:01:38,320 --> 00:01:40,000
performance of the best models is

48
00:01:40,000 --> 00:01:42,880
matching them. Even GBT 5.6 so which is

49
00:01:42,880 --> 00:01:44,960
a pretty capable model and this is set

50
00:01:44,960 --> 00:01:48,000
at max on the artificial analysis index.

51
00:01:48,000 --> 00:01:51,520
This model achieves 61 and Grock 4.6 61.

52
00:01:51,520 --> 00:01:53,840
So it ties it and it's much cheaper. And

53
00:01:53,840 --> 00:01:55,119
then even on all of these other

54
00:01:55,119 --> 00:01:57,439
benchmarks, for example, the GDP Val

55
00:01:57,439 --> 00:01:59,600
one, it actually beats Fable 5, which is

56
00:01:59,600 --> 00:02:04,159
at 1741. And then GPT 5.6, it's at 1728.

57
00:02:04,159 --> 00:02:06,000
I'm not sure why they didn't choose Opus

58
00:02:06,000 --> 00:02:07,520
5 as well, but I guess they wanted to

59
00:02:07,520 --> 00:02:09,759
choose the quote unquote strongest model

60
00:02:09,759 --> 00:02:11,280
lineup from each lab. And they chose

61
00:02:11,280 --> 00:02:13,760
Fable 5 for Enthropic, which is fair.

62
00:02:13,760 --> 00:02:15,440
And then Deep Software Engineering one,

63
00:02:15,440 --> 00:02:17,520
which is a critical benchmark. This

64
00:02:17,520 --> 00:02:20,640
model doesn't beat Fable 5 or GPT 5.6

65
00:02:20,640 --> 00:02:22,480
six soul, but it gets close to it. It's

66
00:02:22,480 --> 00:02:26,080
65.9 and Fable 5 sits at 70%. But if

67
00:02:26,080 --> 00:02:27,520
you're getting results that are pretty

68
00:02:27,520 --> 00:02:29,200
close and the model is five times

69
00:02:29,200 --> 00:02:30,879
cheaper, I wouldn't be too disappointed

70
00:02:30,879 --> 00:02:32,560
with this result. And then same thing

71
00:02:32,560 --> 00:02:34,879
with Cursor Bench 3.2, the model

72
00:02:34,879 --> 00:02:36,959
achieves 69.9%,

73
00:02:36,959 --> 00:02:40,080
funny number. And Fable 5 sits at 70.5.

74
00:02:40,080 --> 00:02:42,400
So once again, closer. And Grock 4.6

75
00:02:42,400 --> 00:02:45,200
beats GPT 5.6 Soul. Same thing with the

76
00:02:45,200 --> 00:02:48,000
Frontier Code. It gets close to Fable 5,

77
00:02:48,000 --> 00:02:51,040
beats GPT 5.6 Soul. So yeah, this model

78
00:02:51,040 --> 00:02:53,280
is actually available in cursor. So the

79
00:02:53,280 --> 00:02:55,200
partnership with cursor or I guess the

80
00:02:55,200 --> 00:02:57,360
acquisition has really helped SpaceX

81
00:02:57,360 --> 00:02:59,360
make some strides in the AI space this

82
00:02:59,360 --> 00:03:01,840
year and Grock build is something that

83
00:03:01,840 --> 00:03:03,680
you know maybe not a lot of us have been

84
00:03:03,680 --> 00:03:05,599
using so far but it's probably going to

85
00:03:05,599 --> 00:03:08,159
be another platform like codeex or cloud

86
00:03:08,159 --> 00:03:10,319
code that we start to use but obviously

87
00:03:10,319 --> 00:03:12,319
cursor is quite strong as well. So you

88
00:03:12,319 --> 00:03:14,319
have options available for you to use

89
00:03:14,319 --> 00:03:16,319
them in both. And one thing to note is

90
00:03:16,319 --> 00:03:18,560
that they're offering two times usage

91
00:03:18,560 --> 00:03:20,560
inside Grock build and cursor for the

92
00:03:20,560 --> 00:03:22,239
first week. So if you just want to try

93
00:03:22,239 --> 00:03:24,640
it out, see what you feel about it, then

94
00:03:24,640 --> 00:03:26,400
you know it might be worth trying it out

95
00:03:26,400 --> 00:03:28,000
right now cuz you get double the usage

96
00:03:28,000 --> 00:03:30,000
in the first week. Now if we were to

97
00:03:30,000 --> 00:03:31,519
take a look at some of the outputs that

98
00:03:31,519 --> 00:03:33,280
people have been generating with Grock

99
00:03:33,280 --> 00:03:35,519
4.6, what we're looking at right now is

100
00:03:35,519 --> 00:03:38,080
a Falcon 9 booster return sequence

101
00:03:38,080 --> 00:03:40,000
simulation. And this was done in a

102
00:03:40,000 --> 00:03:43,360
single HTML file. And as I said, if you

103
00:03:43,360 --> 00:03:45,360
expected to get this type of output from

104
00:03:45,360 --> 00:03:48,560
Grock in 2025, you would be kind of

105
00:03:48,560 --> 00:03:50,080
surprised because you wouldn't expect

106
00:03:50,080 --> 00:03:51,680
something like this to be generated with

107
00:03:51,680 --> 00:03:53,360
Grock. But now it looks like we have to

108
00:03:53,360 --> 00:03:55,120
start taking the Grock team a little bit

109
00:03:55,120 --> 00:03:57,360
more serious because this output is

110
00:03:57,360 --> 00:03:59,360
quite competitive. And obviously, we're

111
00:03:59,360 --> 00:04:01,120
just looking at a simulation and we're

112
00:04:01,120 --> 00:04:02,879
just basing it off of a visual

113
00:04:02,879 --> 00:04:04,959
representation. But if the model is able

114
00:04:04,959 --> 00:04:06,080
to produce something like this

115
00:04:06,080 --> 00:04:08,400
consistently, then I would expect a lot

116
00:04:08,400 --> 00:04:10,159
of people to start adopting Grock

117
00:04:10,159 --> 00:04:11,840
because number one, it is cheaper than

118
00:04:11,840 --> 00:04:13,680
the other labs at least at the moment

119
00:04:13,680 --> 00:04:15,120
because we don't know if this pricing

120
00:04:15,120 --> 00:04:17,040
strategy is going to be sustainable for

121
00:04:17,040 --> 00:04:19,040
the SpaceX team in the long run. But at

122
00:04:19,040 --> 00:04:20,560
least for now, their models are

123
00:04:20,560 --> 00:04:22,079
definitely cheaper compared to the

124
00:04:22,079 --> 00:04:24,160
others. And as I mentioned, this model

125
00:04:24,160 --> 00:04:26,160
excelled at the artificial analysis

126
00:04:26,160 --> 00:04:28,800
index. This model jumped to probably

127
00:04:28,800 --> 00:04:32,720
number four model. It's tied to GPT 5.6

128
00:04:32,720 --> 00:04:34,000
pretty much similar. So you could say

129
00:04:34,000 --> 00:04:35,759
number three as well, but the models

130
00:04:35,759 --> 00:04:38,639
before that are Fable 5 and Opus 5,

131
00:04:38,639 --> 00:04:40,720
which are only above the model by about

132
00:04:40,720 --> 00:04:43,840
a 1% or a 2% difference. So yeah, even

133
00:04:43,840 --> 00:04:46,000
on this intelligence index, which if

134
00:04:46,000 --> 00:04:47,520
you're not familiar with has nine

135
00:04:47,520 --> 00:04:49,360
evaluations. So on all of these

136
00:04:49,360 --> 00:04:51,440
evaluations, it's kind of matching

137
00:04:51,440 --> 00:04:53,600
almost Fable 5 performance, which is

138
00:04:53,600 --> 00:04:56,240
crazy to see. Before we continue, we

139
00:04:56,240 --> 00:04:57,840
just launched the Universe of AI

140
00:04:57,840 --> 00:04:59,600
newsletter. If you want to stay on top

141
00:04:59,600 --> 00:05:01,600
of AI news without having to hunt for

142
00:05:01,600 --> 00:05:03,680
it, link is in the description. Don't

143
00:05:03,680 --> 00:05:05,520
miss out. And what you see on screen

144
00:05:05,520 --> 00:05:07,680
right now is a racing game that Grock

145
00:05:07,680 --> 00:05:10,400
4.6 build. And based off of this post,

146
00:05:10,400 --> 00:05:12,320
the model took about 1 minute and it was

147
00:05:12,320 --> 00:05:14,400
a fiveword prompt, which was a create a

148
00:05:14,400 --> 00:05:16,639
simple racing game in HTML. So, if

149
00:05:16,639 --> 00:05:17,919
you're able to generate something like

150
00:05:17,919 --> 00:05:20,960
this easily using Grock 4.6, I think a

151
00:05:20,960 --> 00:05:22,880
lot of people will be happy. And this is

152
00:05:22,880 --> 00:05:25,199
a more detailed analysis of what it

153
00:05:25,199 --> 00:05:27,919
costs to run GPT 5.6 6 on the artificial

154
00:05:27,919 --> 00:05:29,919
analysis index and what it produced

155
00:05:29,919 --> 00:05:32,000
meaning the output. Both of these models

156
00:05:32,000 --> 00:05:34,160
if you remember scored 61 on the

157
00:05:34,160 --> 00:05:36,240
artificial analysis index. Now to run

158
00:05:36,240 --> 00:05:38,880
the whole test with GPT 5.6 it cost

159
00:05:38,880 --> 00:05:42,400
about 2.8K and then with Grock 4.6 it

160
00:05:42,400 --> 00:05:45,360
cost about 1.1Kish. And this tells you

161
00:05:45,360 --> 00:05:46,960
that you're getting similar level of

162
00:05:46,960 --> 00:05:49,199
performance at half the cost. So yeah,

163
00:05:49,199 --> 00:05:51,120
this is a big release for the SpaceX

164
00:05:51,120 --> 00:05:52,639
team because they just proved once again

165
00:05:52,639 --> 00:05:54,400
that they are a lab that you seriously

166
00:05:54,400 --> 00:05:56,320
start need to considering, especially in

167
00:05:56,320 --> 00:05:58,560
2026. And I'm going to talk more about

168
00:05:58,560 --> 00:06:01,120
Deep Seek version 4 Pro GA. But

169
00:06:01,120 --> 00:06:02,880
basically what we're seeing today is

170
00:06:02,880 --> 00:06:04,639
that both of these releases kind of

171
00:06:04,639 --> 00:06:07,360
emphasize the fact that performance and

172
00:06:07,360 --> 00:06:09,199
all above that is the price at what

173
00:06:09,199 --> 00:06:10,639
you're getting for that performance is

174
00:06:10,639 --> 00:06:12,479
becoming more and more important for all

175
00:06:12,479 --> 00:06:14,800
users because we see many labs now

176
00:06:14,800 --> 00:06:17,039
focusing on creating the best model at

177
00:06:17,039 --> 00:06:19,600
the cheapest cost. Last year in 2025

178
00:06:19,600 --> 00:06:21,680
most of the labs were just focused on I

179
00:06:21,680 --> 00:06:24,400
would say creating the strongest model.

180
00:06:24,400 --> 00:06:26,080
Yes, cost was important, but I think

181
00:06:26,080 --> 00:06:27,840
most of the times the frontier labs,

182
00:06:27,840 --> 00:06:30,080
meaning OpenAI, Enthropic, were kind of

183
00:06:30,080 --> 00:06:31,759
more lenient on that fact because they

184
00:06:31,759 --> 00:06:34,000
didn't have as strong of a competition.

185
00:06:34,000 --> 00:06:36,479
Intelligent models that are maybe not

186
00:06:36,479 --> 00:06:39,039
always ahead of OpenAI Enthropic, but

187
00:06:39,039 --> 00:06:40,720
match their performance at a fraction of

188
00:06:40,720 --> 00:06:42,800
the cost. So yes, price toerformance

189
00:06:42,800 --> 00:06:45,680
ratio is becoming a critical I would say

190
00:06:45,680 --> 00:06:48,639
indicator in 2026. Now, this is the

191
00:06:48,639 --> 00:06:50,479
updated benchmark chart after the

192
00:06:50,479 --> 00:06:52,240
release of the new model. And one thing

193
00:06:52,240 --> 00:06:53,919
you'll see across the board is that it

194
00:06:53,919 --> 00:06:55,919
matches the top level performance of

195
00:06:55,919 --> 00:06:57,360
many of the models. The one thing

196
00:06:57,360 --> 00:06:58,479
interesting over here is that they

197
00:06:58,479 --> 00:07:00,639
haven't put Opus 5 here for some reason.

198
00:07:00,639 --> 00:07:02,240
There is Fable 5 here that we can

199
00:07:02,240 --> 00:07:04,000
compare this model against. But one

200
00:07:04,000 --> 00:07:06,000
thing you'll notice is that Deep Seek,

201
00:07:06,000 --> 00:07:09,199
remember this model costs 43.5 cents per

202
00:07:09,199 --> 00:07:12,080
million input tokens and 87 cents per

203
00:07:12,080 --> 00:07:13,919
million output tokens. While the other

204
00:07:13,919 --> 00:07:15,680
models all over here are way more

205
00:07:15,680 --> 00:07:17,440
expensive than that. So the first thing

206
00:07:17,440 --> 00:07:20,000
if you look at terminal bench 2.1 the

207
00:07:20,000 --> 00:07:22,319
model scores 87.9.

208
00:07:22,319 --> 00:07:24,960
The older version of the model was 72.1

209
00:07:24,960 --> 00:07:27,120
and the flash version which we got last

210
00:07:27,120 --> 00:07:30,319
week was 82.7. And what's crazy is that

211
00:07:30,319 --> 00:07:34,000
Fable 5 is 88. Yes, 88. So this model is

212
00:07:34,000 --> 00:07:37,360
only.1% behind Fable 5. And then on the

213
00:07:37,360 --> 00:07:39,520
Cyber Gym, which is Cyber Security, the

214
00:07:39,520 --> 00:07:42,240
model actually beats Fable 5. Fable 5

215
00:07:42,240 --> 00:07:45,280
sits at 83.1 while Deepseek version 4

216
00:07:45,280 --> 00:07:47,039
Pro, the new one that we got today, is

217
00:07:47,039 --> 00:07:49,680
at 83.3. Now, if this tells you

218
00:07:49,680 --> 00:07:51,680
something is that this model is once

219
00:07:51,680 --> 00:07:53,759
again really geared at cyber security.

220
00:07:53,759 --> 00:07:55,599
We saw the Flash model also be geared

221
00:07:55,599 --> 00:07:57,280
towards cyber security and becoming a

222
00:07:57,280 --> 00:07:58,800
model that was quite capable in that

223
00:07:58,800 --> 00:08:00,560
area. And we're seeing the same thing

224
00:08:00,560 --> 00:08:02,720
today with the new version 4 Pro which

225
00:08:02,720 --> 00:08:05,840
sits at 83.3 outperforming Fable 5 on

226
00:08:05,840 --> 00:08:07,680
the benchmark. On deep software

227
00:08:07,680 --> 00:08:09,919
engineering, this model is not beating

228
00:08:09,919 --> 00:08:12,639
Fable 5 because Fable 5 sits at 70, but

229
00:08:12,639 --> 00:08:15,440
the model achieves 62.7 which is ahead

230
00:08:15,440 --> 00:08:18,639
of GLM 5.2 and a bit behind Kimmy K3

231
00:08:18,639 --> 00:08:20,879
which sits at 67.5.

232
00:08:20,879 --> 00:08:22,879
But one thing again, this is a big jump

233
00:08:22,879 --> 00:08:24,560
compared to the preview version of the

234
00:08:24,560 --> 00:08:27,360
model which was at 12.8. So yeah, they

235
00:08:27,360 --> 00:08:28,960
have really trained this model and they

236
00:08:28,960 --> 00:08:30,560
have improved it on the back end because

237
00:08:30,560 --> 00:08:32,880
we can clearly see in this deep software

238
00:08:32,880 --> 00:08:34,880
engineering benchmark. And there's a

239
00:08:34,880 --> 00:08:36,719
couple of other benchmarks, but one key

240
00:08:36,719 --> 00:08:39,120
benchmark that this model excels at is

241
00:08:39,120 --> 00:08:41,279
the automation bench where the model

242
00:08:41,279 --> 00:08:45,760
achieves 31.8 and Fable 5 29.1. So once

243
00:08:45,760 --> 00:08:48,000
again, the model outperforms Fable 5.

244
00:08:48,000 --> 00:08:49,920
Now this is a fraction of the cost as I

245
00:08:49,920 --> 00:08:52,480
mentioned. This is 43 cents and 87 for

246
00:08:52,480 --> 00:08:54,640
input and output respectively versus

247
00:08:54,640 --> 00:08:57,519
Fable 5 which is at $10.50.

248
00:08:57,519 --> 00:08:59,279
So yes, if we look at the terminal bench

249
00:08:59,279 --> 00:09:01,279
for example, we are seeing this model

250
00:09:01,279 --> 00:09:03,600
achieve a result which is 0.1 behind the

251
00:09:03,600 --> 00:09:06,160
best model out there Fable 5 and the

252
00:09:06,160 --> 00:09:09,519
pricing is insane to look at $43 versus

253
00:09:09,519 --> 00:09:13,519
$10 87 versus $50 per million input

254
00:09:13,519 --> 00:09:15,360
tokens. So when we turn that into a

255
00:09:15,360 --> 00:09:18,160
fraction, this model is 57 times

256
00:09:18,160 --> 00:09:20,399
cheaper. And earlier last week, I think

257
00:09:20,399 --> 00:09:22,160
I made a video about how DeepSeek

258
00:09:22,160 --> 00:09:24,480
version 4 Pro was going to achieve a

259
00:09:24,480 --> 00:09:25,839
result like this because there was a

260
00:09:25,839 --> 00:09:27,839
investor report that leaked. And when I

261
00:09:27,839 --> 00:09:29,440
saw the pricing where it said that it

262
00:09:29,440 --> 00:09:30,959
was going to match Fable 5 or even

263
00:09:30,959 --> 00:09:33,760
outperform it and be at 57 times cheaper

264
00:09:33,760 --> 00:09:35,440
per cost. When I was reading the

265
00:09:35,440 --> 00:09:37,120
investor report, I didn't really believe

266
00:09:37,120 --> 00:09:39,040
it. But today, it is proven because at

267
00:09:39,040 --> 00:09:41,040
least on the benchmark so far, yes, I'm

268
00:09:41,040 --> 00:09:42,880
just saying the benchmarks, we are

269
00:09:42,880 --> 00:09:45,040
seeing similar level performance. Now,

270
00:09:45,040 --> 00:09:46,959
if this model actually performs like

271
00:09:46,959 --> 00:09:48,959
that in production, it's a little too

272
00:09:48,959 --> 00:09:50,720
early to tell yet. We still would have

273
00:09:50,720 --> 00:09:52,399
to give it a couple of weeks and see how

274
00:09:52,399 --> 00:09:54,399
it's performing in the long run because

275
00:09:54,399 --> 00:09:56,399
sometimes on the release day the models

276
00:09:56,399 --> 00:09:58,480
perform quite good, but over time they

277
00:09:58,480 --> 00:10:00,240
kind of deteriorate in their quality.

278
00:10:00,240 --> 00:10:02,000
So, we hope DeepS doesn't do that.

279
00:10:02,000 --> 00:10:03,680
Historically, they haven't done that,

280
00:10:03,680 --> 00:10:05,600
but let's just see. Just going to put it

281
00:10:05,600 --> 00:10:07,120
out there because right now we're basing

282
00:10:07,120 --> 00:10:09,760
this off of benchmarks. But that's it

283
00:10:09,760 --> 00:10:11,519
for today's video. Make sure you guys

284
00:10:11,519 --> 00:10:13,519
are subscribed to the channel. Follow

285
00:10:13,519 --> 00:10:15,120
our new newsletter as well at

286
00:10:15,120 --> 00:10:17,350
universeofai.behive.com

287
00:10:17,350 --> 00:10:17,360
universeofai.behive.com

288
00:10:17,360 --> 00:10:19,040
as well as subscribe to the main channel

289
00:10:19,040 --> 00:10:21,519
World of AI and support us on X by

290
00:10:21,519 --> 00:10:23,760
following the Universe of AIZ as well.

291
00:10:23,760 --> 00:10:25,360
Until then, I'll see you guys in the

292
00:10:25,360 --> 00:10:27,519
next