Qwen3-Max Thinking beats Gemini 3 Pro and GPT-5.2 on Humanity's Last Exam (with search)

On HMMT Feb 25, a rigorous reasoning benchmark, Qwen3-Max-Thinking scored 98.0, edging out Gemini 3 Pro (97.5) and ...
MUO on MSN

I don’t need Perplexity anymore because my local LLM does it better

Perplexity was great—until my local LLM made it feel unnecessary ...

DeepSeek AI search is the clearest sign it wants Google’s turf

DeepSeek is hiring for DeepSeek AI search, a multilingual, multimodal engine that could challenge Google’s search habit. The ...
scmp.com

DeepSeek kicks off 2026 with paper signalling push to train bigger models for less

Chinese artificial intelligence start-up DeepSeek has ushered in 2026 with a new technical paper, co-authored by founder Liang Wenfeng, that proposes a rethink of the fundamental architecture used to ...
CNX Software

Raspberry Pi AI HAT+ 2 review – A 40 TOPS AI accelerator tested with Computer Vision, LLM, and VLM workloads

Raspberry Pi sent me a sample of their AI HAT+ 2 generative AI accelerator based on Hailo-10H for review. The 40 TOPS AI ...
WinBuzzer

DeepSeek Targets Google with Multimodal AI Search

DeepSeek has unveiled plans for a multimodal AI search engine processing text, images, and audio, challenging Google's ...

How Google’s Lang Extract Turns Messy Documents into Trustworthy JSON and Interactive HTML

Google’s Lang Extract uses prompts with Gemini or GPT, works locally or in the cloud, and helps you ship reliable, traceable data faster.
Hosted on MSN

DeepSeek blew up markets a year ago. Why hasn't it done so since?

AI stocks went into a freefall in January as relatively unknown Chinese player DeepSeek unveiled a model to rival the world's best. A year on, the AI company's more recent releases haven't caused the ...