DeepSeek unveils DSpark for 60% to 85% faster inference optimization

DeepSeek open-sourced DSpark, a speculative decoding framework that attaches a draft module to existing DeepSeek-V4 weights. It pairs a parallel draft backbone with a lightweight Markov head to cut suffix decay, then adds confidence-scheduled verification that tailors how many tokens get checked to real-time GPU load. Offline, accepted length rises 16–31% over DFlash and Eagle3; in production it speeds per-user generation 57–85% over the MTP-1 baseline, losslessly. The training repo, DeepSpec, ships under MIT. The post DeepSeek Releases DSpark, a Speculative Decoding Framework That Accelerates DeepSeek-V4 Per-User Generation 60–85% Over MTP-1 appeared first on MarkTechPost.

A team cut their AI inference bill by more than half. Three months later, customer satisfaction was dropping and the cost savings were tied to the quality loss. Cost-optimization routing layers are a Pareto trap, and here's the detection methodology that catches them in days instead of months. The post We Built a Routing Layer to Cut Our AI Costs. It Broke the Product. appeared first on Towards Data Science.

Chinese tech giant Tencent is set to launch an AI assistant inside WeCom, its Slack-like collaboration tool for enterprises. The new tool, Dayuan, is built on the latest large language models from Chinese AI developer DeepSeek. Tencent announced the news in a post on Chinese messaging platform Weibo by Tencent’s public relations manager Zhang Jun. Dayuan will automatically understand user requests and will respond according to the demands of the user, he wrote, according to a translation by Bloomberg. “At any time within WeCom, simply swipe left to summon Dayuan. It can intelligently recognize the interface you’re on, understand what you’re asking, and help you resolve issues more effectively,” he wrote, according to the report. In addressing the Chinese enterprise market, Tencent has an advantage over other companies in the AI space because it has a vast reservoir of customers who use WeCom. Earlier this month, it announced a range of AI productivity agents to address the demand for m

Unconventional AI, a startup founded by Naveen Rao, formerly head of AI at Databricks, has released its first AI model and an accompanying research paper detailing a radical reimagining of computing architecture designed to dramatically reduce the energy cost of AI inference. The company’s debut model, Un-0, is an image-generation system built on an oscillator-based […]

Advertised roles suggest company focused on commercialising frontier research

DeepSeek's funding and expansion could intensify AI competition, emphasizing talent retention and efficient model development in the tech sector. The post DeepSeek plans to double staff after raising $7.4 billion in first external funding round appeared first on Crypto Briefing.

AI inference company Groq has closed a $650 million funding round as it pivots its business following a landmark IP licensing agreement with Nvidia. The round was led by Disruptive, a Dallas-based late-stage investment firm whose founder Alex Davis also serves as Groq’s chairman, alongside Fort Lauderdale hedge fund Infinitum. The raise comes roughly six […]

OpenAI has just revealed a new "intelligence processor" chip for AI servers made in partnership with Broadcom. The chip, called Jalapeño, is designed to power current and future large language models, according to an announcement on Wednesday. Jalapeño is an ASIC (Application-Specific Integrated Circuit), meaning it's designed for a specific purpose: AI inference. With AI inference, models process a user's request to run an agent like Codex or offer a response from ChatGPT, while AI training involves a model consuming vast amounts of data to inform its responses. It comes just nine months after OpenAI revealed that it would team up with Br … Read the full story at The Verge.