Google also released a new version of Gemma 3, optimized with ‘Quantization-Aware Training’ This reduces the 27B model's memory requirements, enabling it to run on consumer GPUs, like the NVIDIA RTX 3090, with maintained performance
BIG TECH
-
Google Releases Gemini 2.5 Flash Hybrid Reasoning AI Model
By
–
Google continues to cook—this time dropping Gemini 2.5 Flash, a hybrid reasoner AI —Matches 04-mini
—Strong across reasoning, STEM, and visual reasoning
—Includes 'thinking budget' to optimize for cost
—Available via Google AI Studio, Vertex, Gemini app -
Google Releases New Gemini Model with Thinking Budget
By
–
TODAY'S AI NEWS: Google just released a new Gemini model with "thinking budget." Plus, more news from OpenAI, Alibaba, biotech company Profluent, and DeepMind. Here's everything you need to know:
-
OpenAI vs LangGraph: The Silent War in Agent Engineering
By
–
The Great Debate: OpenAI vs LangGraph https://
latent.space/p/oai-v-langgr
aph
… The silent war in Agent Engineering gets loud. -
ChatGPT o3 and Gemini Deep Search surpass Perplexity
By
–
ChatGPT o3 / Gemini Deep Search > Perplexity
-
Smart People and Compute Resources Will Ensure Success
By
–
They will be fine, lots of smart people and compute there
-
Whistleblower Case Against Tech Company Lacks Valid Excuse
By
–
The excuse of being busy doesn't hold up: he has a large team handling many cases, he already made the time and commitment, it's a highly relevant whistleblower for many of his cases, it's a company named as defendant in their complaints too.
-
Strategic Naming Mystery Behind OpenAI’s Model Versions
By
–
It may have been an accident at first, but now it does seem strategic. It must be. What does the "o" even stand for? They implied for 4o it was "omni" (image + text) but then what does it stand for in o4? There's no good answer. Which means 4D chess is the only logical reason.
-
TPU Hardware Architecture: From v1 to v4 Optical Interconnect
By
–
If you want to read about them from the hardware perspective, read these three papers: https://
arxiv.org/abs/1704.04760
(about TPU v1, the one that started it off) https://
arxiv.org/abs/2304.01433 (about TPU v4 and its optical interconnect) https://
gwern.net/doc/ai/scaling
/hardware/2021-jouppi.pdf
… (ten lessons) -
AI Model Naming Strategy: Intentional Annoyance for Free Marketing
By
–
One day we'll find out they named their AI models terribly on purpose. That the tons of free PR was the goal all along. "What's the most irritating way we could possibly name these?" "What if we just put the "o" on either side of the number… like 4o or o4." "Oh, love that."