My article on Rust is trending on HackerNews, and this time I didn't post it! Would anyone be interested in a video series on this? karimjedda.com/carefully-exp…

🚀 Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable. 🔗 github.com/deepseek-ai/DeepI… 1/4

🚀 DeepSeek-V4.1-Flash-1017 is here! Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API. 🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model. 1/4

Should Europe leave frontier AI to the Americans and the Chinese? The idea has gained ground in Europe. It is a fundamental mistake. Build or surrender: for now, the choice is still ours. Full article below.

Muse Spark Open -- weights releasing today -- runs great on new Mac Studios thanks to work with @johnternus and Apple.

Meet Mistral Large 4, aka Le Chonk. • 1T parameters, natively multimodal. 49B active. It is the best open weights model from US or Europe on aggregated benchmarks. • State-of-the-art on critical workloads, including cyber defense, manufacturing and finance and it surpasses closed frontier models on visual grounding. • Forged in Europe end-to-end and is deployable from Europe via our own Mistral Cloud infrastructure. • Available to all via API today. Working with cybersecurity partners privately. Open weights release end of October.

25 engineers ran a self-hosted open-weight model for one month, generating 13 billion tokens. But our trial continued running because people kept using it, unprompted. Along the way, we uncovered real insights into owning our own AI infrastructure: cost, availability, model choice, data, and future capabilities, instead of depending entirely on external providers. parity.io/blog/self-hosted-a…

Self-hosted open-weight / open-source AI is very viable, we now have data. Thanks to everyone advancing the "frontier" and specifically thanks to the @deepseek_ai for releasing such amazing models which we tested here. We'll test more!

Today is the 21st of September.

I'm very sensitive to this feeling, cause I've been there. Someone does something cool, gets a lot of attention, and you're frustrated cause you already did it, a while ago. It's not enough to do things, you must also tell people. Maybe it's not fair, but it's true

Did you know tiny models are capable of out-planning a database? I trained a 4B model to produce 81% faster query plans than Postgres. Write-up below 👇

Almost 7 years apart, Cumulus is taking it next step to #JAM. Parachain service is doing its first steps. @Polkadot

🚀 Introducing DeepSeek-V4.1-Flash: smarter, faster, more efficient. 🔹 Introducing the smallest model in our new architecture family, with native visual understanding. 🔹 Designed for greater capability, faster inference, higher throughput, and scaling to larger models. 1/6

The OpenDisplay macOS Receiver app just dropped. It supports mac back to macOS 12!

The new AI model that’s swept to the top of online usage charts with high performance at zero cost was created by China’s Z.AI, also known as Zhipu bloomberg.com/news/articles/…

🚨I've just released Sentence Transformers v6.0! MultiVectorEncoder joins the family: ColBERT-style late interaction models are now a first-class model type, for training, inference & interpretation, alongside dense, sparse & reranker models. Big thread 🧵

Qwen3.8 27B brings a new state-of-the-art dense model for local AI development. ⚡ Run it on AMD Ryzen™ AI Max+ processors or single Radeon™ AI PRO R9700 card ⚡ Experience it with @LMStudio ⚡ Turn it into an app with @lemonade_server Start building with AMD Day 0 support for @Alibaba_Qwen now: amd.com/en/blogs/2026/run-qw…

We promised open weights for Qwen3.8. Now, time to meet them! 🎉 ⚡ Qwen3.8-27B: - A native multimodal dense model. With just 27B parameters, it outperforms Qwen3.7-Plus overall and shines in real-world coding & office workflows. - 262K native context, easily extendable to 1M tokens via YaRN. - Built for builders. Highly efficient, high-quality, and licensed under Apache 2.0. 🚀 The open weights for Qwen3.8-2.4T-A95B (Max-level) have also been released recently. Whether you're shipping lightweight applications with Qwen3.8-27B locally or building agents with Qwen3.8-2.4T-A95B, they're yours now! Download, deploy, and build something we haven't imagined yet. 👀👇 - Hugging Face： huggingface.co/collections/Q… - ModelScope： modelscope.cn/collections/Qw…

I don't think the DeepSeek Harness is perfect but this is for sure the first time I have been looking at something new in the space and felt quite inspired to revisit some of our choices. I love that part about Open Source a lot!