Carlos Jiménez

Artificial Intelligence Researcher 🐍🇪🇨

[Shruti @heyshrutimishra]

🚨 BREAKING: Google just turned Gemini into an AI operating system.

At I/O 2025, they launched a full suite of tools that go way beyond chat.

Here’s what just dropped and it’s wild: 🧵👇

[Carlos Jiménez @CaJiFan]

First phase ends next week, hurry!!

#CVPR2025 #MetaFood #VLMs
#FoundationModels #Challenge

[Bhalaji @ CVPR2026 @bhalaji10]

🫠 Between deadlines, rebuttals & existential crises??? "We got you!!!"
📢 Our @CVPR MetaFood 2025 Multi-Modal Challenge is now LIVE!!!
🍽️ Dishcovery: VLM MetaFood Challenge 🍽️
🧰 Task: Match fine-grained food images to detailed descriptions.
🔗 Join us: https://t.co/AOerM7btUF

[Andrew Ng @AndrewYNg]

It finally happened -- thanks to people learning to write AI code, Python is now the top programming language on GitHub!

If you want to learn Python, check out https://t.co/zpIxRSuky4's free course AI Python for Beginners.

[Carlos Jiménez @CaJiFan]

@mgonzaalez_ Diablo

[Josep Ferrer @iamjosepferrer]

The Transformer's encoder clearly explained 👇🏻

[El Profeta @EiProfeta]

La escena eliminada que anunciaba la muerte de Albus Dumbledore. Increíble.

[Carlos Jiménez @CaJiFan]

Funca

[Hernando Abella @hernandoabella]

😂😂

[Carlos Jiménez @CaJiFan]

Mood

[Carlos Jiménez @CaJiFan]

De programación

[Ana Villanueva 🐄 @ana_gvillanueva]

si

[El Profeta @EiProfeta]

Hace 27 años, era publicado "Harry Potter y la piedra filosofal". El comienzo de todo.

[El Profeta @EiProfeta]

Si soy.

[Carlos Jiménez @CaJiFan]

Va por el buen camino dice JAJJAA

[alejajandro @alejajandrooo]

Como de jodido estoy

[Ryan Els @RyanEls4]

An acceptable distraction while coding 😂😊

[S4vitar @S4vitar]

Y recuerda, nunca te despedirán del trabajo si acostumbras a no comentar tu código y además eres el único que sabe cómo funciona.

[Carlos Jiménez @CaJiFan]

@mgonzaalez_ @Genesispbbbbb Confirma

[Carlos Jiménez @CaJiFan]

@mgonzaalez_ El futuro es hoy

[Carlos Jiménez @CaJiFan]

We expect these Meta/Google/OpenAI models to drop like they’re some kind of fresh rap album 😂

[Bindu Reddy @bindureddy]

Key Evals For GPT-4o

There are two ways to read these evals.

Option 1 - Is this the mind-blowing new model that was supposed to be a huge leap forward from GPT-4. No! It's not

In fact, it proves that LLMs are plateauing and that each extra point gained in MMLU is very hard to achieve. OpenAI's choice to name it GPT-4o says it all!!

Option 2 - Is this one of the best models based on benchmarks? Their published MMLU and Human Eval look better than their current best model. I suspect Llama-3 400B can beat or match it, but it's not been released. So, technically, GPT-4o is the best model in the world.

Until we hear from Google or Meta drops the 400B model.