Deepseek's latest AI model is smarter than ever, but its low running costs come at a price

Chinese AI lab releases open-weight model engineered for low-cost agentic inference

By GPU Data Hub Desk · GPU Data Hub·Editorial standards·23 Oct 2026, 12:00·China

Deepseek's latest AI model is smarter than ever, but its low running costs come at a price

DeepSeek has launched its 4.1 Flash 1017 artificial intelligence model, targeting agentic automation with sharply reduced inference expenses. The open-weight release aims to drive task-level computing costs down further while maintaining competitive operational capability. However, the architecture's ultra-low operational budget reportedly involves performance trade-offs under certain processing workloads.

Original source

This summary was written by the GPU Data Hub desk from reporting published by Tom's Hardware on 23 Oct 2026, 12:00. Read the full article at the original publisher.

Read Original Source →

Companies

DeepSeek

Topics

open-weight models inference efficiency agentic AI compute costs

LinkedIn X

More from GPU Data Hub

Latest →

Meta delays new business terms after publisher pushback on AI scraping

Meta has postponed the introduction of revised commercial terms to 20 November following resistance from media organisations. Content creators and publishers expressed apprehension that the updated conditions would authorise the tech group to harvest web content for artificial intelligence training.

AI Infrastructure·3h ago·The Verge

Nvidia's Huang heads to Kazakhstan for the first time — and he's not the only one with big plans

Nvidia chief executive Jensen Huang has travelled to Kazakhstan for the first time, underscoring the country's accelerating efforts to establish itself in the global artificial intelligence landscape. The engagement illustrates how major compute providers are exploring sovereign infrastructure opportunities in previously non-traditional technology hubs.

AI Infrastructure·8h ago·CNBC Technology

Vendor-affiliated AI agent tech gets adopted by friendly, funded foundation

A well-resourced foundation has assumed stewardship of four software tools engineered to advance artificial intelligence agents. The transition moves the software away from proprietary vendor influence and into an independent administrative framework.

AI Infrastructure·10h ago·The Register

Uber rival inDrive eyes 2027 robotaxi tests in Central Asia with Pony.ai and Nvidia

Ride-hailing operator inDrive is preparing to launch autonomous vehicle trials in the capital of Kazakhstan in the first half of 2027. The project will deploy self-driving vehicles developed by Pony.ai, drawing on hardware from computing firm Nvidia.

AI Infrastructure·13h ago·TechCrunch

Kazakhstan Expands Compute for Its AI Future

Kazakhstan is transitioning its national artificial intelligence initiative from preliminary roadmaps into live implementation across the entire computing stack. The country is ramping up its accelerated computing footprint to support rising processing demands spanning foundational models and software deployments.

AI Infrastructure·15h ago·NVIDIA Newsroom

Europe RAISEs the Stakes for AI in Science

The European Commission has broadened its Resource for AI Science in Europe scheme, introducing the AI In Science Enablers programme to attract external backing and capital. European authorities have also established three supplementary resources, comprising the RAISE Hub and the RAISE Forum, to assist scientific computing endeavours.

AI Infrastructure·19h ago·HPCwire