DeepSeek just open-sourced its secret sauce
DeepSeek's new chip on the block isn't Nvidia
DeepSeek has open-sourced the inference engine it uses to serve its models in production, a step it said last year wasn’t yet practical.
Dubbed DeepInfer, the framework is designed to let developers run DeepSeek’s V4.1 Flash model on their own infrastructure, with many of the same optimizations the company uses behind its API. Support for future DeepSeek models is also planned.
DeepSeek has, of course, long made its model weights available for developers to download and run using third-party software such as vLLM. But running a model efficiently is a different matter altogether. The software that manages requests, distributes work across chips, and generates responses can have a significant impact on speed, throughput and computing costs. And so DeepInfer essentially gives developers access to DeepSeek’s own approach to those challenges.
More fresh reads from TNS:
▹ElevenLabs wants to run your voice agent. Here’s what developers give up.
▹MiniMax just halved its token prices. Your AI agent bill may not follow.
▹AI’s Midas problem: When code is free, features are worthless
Join us during KubeCon for a night at the Museum of Illusions
The New Stack is excited to be activating at KubeCon + CloudNativeCon in Salt Lake City November 9-12! We’ll be covering the biggest news, doing interviews, and recording videos. Plus, we’re teaming up with Dynatrace to co-host what promises to be your most memorable evening of the conference. Bend your brain and expand your network by joining us at the Museum of Illusions as we celebrate Kubernetes and the cloud native community.
TOP OF THE STACK
Beam pulls single app windows off the Mac Mini running your agents
More developers now leave a Mac Mini running agents around the clock. Beam lets them reach into it one app window at a time instead of opening a whole second desktop.
Beam is a new Mac app that takes a window running on one Mac and puts it on another, where it behaves like a locally installed app. Why? Agents. Serial founder Hiten Shah, who co-founded Crazy Egg and KISSmetrics, built it with Sam Asante under...
WHAT ELSE IS NEW?
Ship on day one. Accelerate on day two.
Hope builds apps with clean architecture, open dev standards and a complete dev experience on your mobile, tablet and desktop.
- Portable, open, standard code. No vendor lock-in.
- Live staging for every change before production.
- Kubernetes and MongoDB, hosted or on your cloud.
Starter plan free for 3 months, a $90 value ($29.99/mo after). Use code SHIP.
FLOW STATE
Trusting an agent to respond to an incident is one thing. Trusting it to decide when an incident needs attention is another. Join us live on October 22 to see how that shift plays out in production.
Forward Deployed Engineers sit at the intersection of engineering and customers — writing code one day, solving problems on-site the next. This step-by-step roadmap breaks down exactly what skills, tools, and experience you need to land the role in 2026.
AI agents made building internal apps fast. Getting them safe access to live production data is still slow, manual, and scattered across every system.
How did we do today?
ALSO FROM THE IMG FAMILY
More communities worth exploring
A modestly upbeat breakdown of what's happening with tech, business and power.
Community-powered role and skill-based roadmaps to guide your learning journey in tech.
Insights from the brightest minds in data science, machine learning and AI.
Join TNS on your favorite platform
The New Stack is a media platform for the people who build and manage software the world relies on.
© 2026 The New Stack. All rights reserved.
The New Stack 1111 6th Ave Ste 550 PMB 50938 San Diego, CA 92101-5211