LLMS

Claude is claimed to be 3x better at instruction following

By

–

28 August 2025 15h16

claude is 3x better at instruction following for me

User claims Gemini is 5x better and faster

By

–

28 August 2025 15h10

gemini is like 5x better and faster lol

→ View original post on X — @aymericroucher

Evaluating Gemini 2.5 Flash on ARC-AGI benchmarks

By

@aymericroucher

–

28 August 2025 15h05

I've tested if nano-banana / Gemini-2.5-flash-image beat ARC-AGI – it's quite far. Btw bravo to the ARC_AGI team, the delta between easiness of problems for humans vs difficulty for LLMs is just

AI Energy Use: Should Companies Disclose Per-Prompt Costs?

By

–

28 August 2025 11h44

AI isn’t free. Every question we ask consumes power, water, and resources. Google just opened the door to a conversation the entire industry has been avoiding. Do you think AI companies should be required to publish per-prompt energy usage?

AI Efficiency Surges in 2025

By

–

28 August 2025 11h43

The bigger surprise? AI efficiency is skyrocketing. In May 2024, a Gemini prompt used 33x more energy than it does today. Software optimization + hardware improvements are cutting the cost fast.

→ Voir le post original sur X — @vision_ia

Gemini Nano Full Version Outperforms Bridged AI Studio Release

By

@vision_ia

–

28 August 2025 10h58

Attention, si vous avez été déçus par Nano Banana, c’est parce que Google a sorti une version bridée sur AI Studio. Il était bon, mais c'était pas celui de LMARENA… Mais ils ont commencé à le rendre accessible via Gemini… et là, c’est une dinguerie… Ils ont ressorti le

→ View original post on X — @therundownai

Nous Research Introduces Hermes 4 Open Hybrid Reasoning Models

By

@therundownai

–

28 August 2025 8h32

Nous Research introduced Hermes 4, a family of open, hybrid reasoning models Designed to be neutral, the models achieve SOTA on RefusalBench, which tests for willingness to be helpful in scenarios usually disallowed by both closed and open models

→ View original post on X — @alliekmiller

Google Releases Gemini 2.5 Flash Multimodal AI Model

By

@alliekmiller

–

28 August 2025 7h19

Gemini 2.5 flash image preview aka nano-banana

→ View original post on X — @emollick

LLMs Beyond Matrix Multiplication: Understanding Model Capabilities

By

@emollick

–

28 August 2025 7h14

Yes. yes LLMs are not just matrix multiplication but adding that there are non-linear functions as well doesn't really do anything to resolve the central mystery of why these models can do what they do. And here is the source of Wolfram's paragraph: