Geeky Gadgets

Minimax M2.5 Benchmarks : Targets $1 per Hour for 100 Tokens per Second

The Minimax M2.5 is a compact yet capable AI model that combines affordability with competitive performance, as detailed by Sam Witteveen. With a processing cost of just $1 per hour for 100 tokens per ...
Geeky Gadgets

MiniMax M2.5 Uses 10B Active Parameters per Token, Aiming for Cheaper Always-On Agents

The MiniMax M2.5 is an advanced AI model designed to balance high performance with cost efficiency, as explained by Caleb Writes Code. Building on its predecessor, the M2.1, this model incorporates a ...
IEEE

Information-Theoretic Minimax Regret Bounds for Reinforcement Learning based on Duality

Abstract: We study agents acting in an unknown environment where the agent’s goal is to find a robust policy. We consider robust policies as policies that achieve high cumulative rewards for all ...
Finimize

MiniMax Heads To Hong Kong For A $700 Million AI IPO

Alibaba-backed MiniMax plans a Hong Kong listing in early January as demand for Chinese AI listings grows, even as the company keeps posting sizable losses. MiniMax has cleared Hong Kong Exchange’s ...
IEEE

A Joint High-Precision Frequency and Code Phase Estimation Algorithm for Iridium NEXT ...

Abstract: In recent years, Low Earth Orbit (LEO) satellite signals of opportunity (SOPs) positioning technology has emerged a significant alternative to the Global Navigation Satellite System (GNSS), ...