The truth is this: no other nation on Earth can match the U.S. when it comes to compute build-out and AI infrastructure. The U.S. owns the vast majority of the world’s advanced chips, the world’s dominant share of usable compute, and (critically!) the ability to network it at
COMPUTING
-
Poetiq beats Gemini 3 on ARC-AGI-2
By
–
Poetiq system achieved 54% score on ARC-AGI-2 benchmark, surpassing Gemini 3 Deep Think at more then twice lower compute cost. Poetiq (Mix) is a combined self learning system that leverages Gemini 3 and GPT-5.1 models. We want to do SVG tests
-
Voyager SDK v1.5.1 Improves Quantised Model Accuracy
By
–
Voyager SDK v1.5.1: Better Quantised Model Accuracy Without Model Changes
— Axelera AI (@AxeleraAI) 5 décembre 2025
The latests version of Voyager SDK quietly boosts quantisation accuracy — no model changes needed. Just recompile your ONNX model and enjoy better FP32 iso-accuracy out of the box. Smarter compiler,… pic.twitter.com/S02GSPkO3OVoyager SDK v1.5.1: Better Quantised Model Accuracy Without Model Changes The latests version of Voyager SDK quietly boosts quantisation accuracy — no model changes needed. Just recompile your ONNX model and enjoy better FP32 iso-accuracy out of the box. Smarter compiler,
-
OpenAI May Release GPT-5.2 Soon
By
–
BREAKING : OpenAI may release GPT-5.2 as early as December 9 according to The Verge. RedGPT
-
Grok 4.20 tested as mystery model
By
–
It turns out that Grok 4.20 is being tested on Alpha Arena as a "mystery model" and shows quite significant performance. "Undisclosed frontier AI lab (not Nof1)" How much would you pay for a model with such performance to run your portfolio?
-
EU Commission and EIB Support AI Gigafactories Initiative
By
–
Commission and European Investment Bank Group team up to support AI Gigafactories | Shaping Europe’s digital future https://
digital-strategy.ec.europa.eu/en/news/commis
sion-and-european-investment-bank-group-team-support-ai-gigafactories
…
#digitaleu #AI #Gigafactories #Innovation #Technology @DigitalEU @ArturHabant @elaniazito @CurieuxExplorer @Shi4Tech @smaksked @BetaMoroney -
AI breakthroughs tackle malnutrition, chips, and accessibility
By
–
Zipline cuts childhood malnutrition.
— Packy McCormick (@packyM) 5 décembre 2025
Dells donate $6.25B to 25M children.
AWS launches Trainium3 chip, titans battle.
Building blocks of life are everywhere.
Michael Levin back on Lex Fridman.
+ Thinking Game ++ Dose Fuel +++ Hearing Aids
What a week for the optimists. pic.twitter.com/gFmsZYcujzZipline cuts childhood malnutrition.
Dells donate $6.25B to 25M children. AWS launches Trainium3 chip, titans battle.
Building blocks of life are everywhere.
Michael Levin back on Lex Fridman. + Thinking Game ++ Dose Fuel +++ Hearing Aids What a week for the optimists. -
6D Parallelism: The Human-like Parallel Processing in AI
By
–
Little-known fact in AI: 6D parallelism is human parallelism
-
Fine-tuning and Serving LLMs: Essential Tools and Frameworks
By
–
Here’s my list: 1. Transformers + TRL – for fine-tuning/ RL on smol LLMs
2. vLLM/ MLX/ llama.cpp for serving them locally – fun for local projects
3. Take it one step further with transformers.js w/ WebGPU backend
4. For remote LLMs w/ agents – OpenAI Agents SDK is pretty dope
