Mostafa Dehghani
@m__dehghani
Followers
9K
Following
11K
Media
147
Statuses
2K
/google_deepmind/gemini/.
San Francisco
Joined June 2014
Clearly this deception eval needs a deception eval!.Come on open ai!.
Hopefully this plot isn't the work of GPT-5… otherwise it needs a patch!
1
1
48
Hopefully this plot isn't the work of GPT-5… otherwise it needs a patch!
5
6
89
Sometimes working on new ideas is just confidently guessing in absolute darkness… until something glows. It's both scary and magical. but also very addictive!.
2
2
102
RT @OriolVinyalsML: Gemini 2.5 Deep Think is rolling out to Google AI Ultra subscribers!. It uses advances in parallel reasoning & RL to ta….
0
38
0
Wanna hang out with a cool math olympiad gold medalist? 🥇.
blog.google
Deep Think utilizes extended, parallel thinking and novel reinforcement learning techniques for significantly improved problem-solving.
5
2
58
The grind is always in crossing the good‑model / good‑system divide.
0
1
10
RT @Mihonarium: 🚨 According to a friend, the IMO asked AI companies not to steal the spotlight from kids and to wait a week after the closi….
0
197
0
Gemini nailed a gold-level score on the IMO (35/42)! 🥇Huge step forward. can’t wait to see what it unlocks for math and science.
Drastic progress on maths with Gemini 2.5! As a math undergrad, I am impressed 🤯. 🥈 -> 🥇 ✅.Formal -> Informal ✅.Specialized model -> General model ✅.Available soon ✅. Huge thanks to IMO and congrats to all participants!. Blog:
0
2
89
RT @AntoineYang2: Thrilled to share our latest advances in video understanding 📽️: Gemini 2.5 Pro is a truly magical model to play with, ex….
developers.googleblog.com
0
51
0
Research failures often come from chasing things that don’t matter long-term & it’s easy to fall into that trap individually, no matter how strong you are. That’s why joining flagship efforts can be great for researchers: the course-correcting power of group intelligence. .
2
9
157
RT @19kaushiks: @OfficialLoganK @m__dehghani @kat_kampf Shoutout to @a7b2_3 and @bigmoonmandybot for driving lots of the improvements here.….
0
2
0
RT @googleaidevs: Image Generation is now available in preview with increased rate limits for Gemini 2.0 Flash. ⚡️
0
78
0
RT @OfficialLoganK: Today we are rolling out an updated Gemini 2.0 Image Generation model with 🖼️:. - Better visual quality .- More accurat….
0
213
0
RT @OfficialLoganK: Read more about todays launch: big congrats to @19kaushiks @m__dehghani @kat_kampf and the res….
developers.googleblog.com
0
10
0
Triple Crowned! 👑👑👑🫡. "This is the first-ever sweep across text, vision, and WebDev by any model!".
🚨Breaking: @GoogleDeepMind’s latest Gemini-2.5-Pro is now ranked #1 across all LMArena leaderboards 🏆. Highlights:.- #1 in all text arenas (Coding, Style Control, Creative Writing, etc).- #1 on the Vision leaderboard with a ~70 pts lead!.- #1 on WebDev Arena, surpassing Claude
0
9
106
Alright, the wait is over. image editing has entered the Gemini chat 🫡.
2
4
64
RT @OfficialLoganK: Native image editing with Gemini is now starting to roll out to all users in the @GeminiApp!!
0
71
0
RT @OfficialLoganK: Gemini 2.5 Pro just got the final 8th badge in Pokemon Blue, incredible pace of progress by the world's most powerful m….
0
223
0
Gemini 2.5 Flash is here! .It’s sharp on code and reasoning, fast, and extremely cost-effective compared to models in its class. Great workhorse for developers. Can’t wait to see what folks build on top it, especially for multimodal! 🫡.
2.5 Flash is a huge jump from 2.0 (which was a huge jump from 1.5)!
1
8
91