Every article we have published, newest first. Search titles and leads, or change the order.
BusinessNews
At GTC Berlin, NVIDIA unveiled DSX POD: factory-built, pre-validated Vera Rubin AI factories in containers. Deployment drops from six months to as little as 14 days. New DSX software components arrived alongside.
· 2 min read
ToolsNews
NVIDIA updated its local AI machines at GTC Berlin: a 64GB DGX Spark, eight-unit clusters, DGX Station for Windows and a new local coding model, Recursive Flint.
· 3 min read
BusinessNews
At GTC Berlin, NVIDIA and Nokia said T-Mobile completed the first AI-RAN call on a commercial network. Deutsche Telekom, BT, Elisa and Orange are starting AI-RAN pilots.
· 2 min read
ToolsNews
At GTC Berlin, NVIDIA made the cuDSS sparse solver library generally available and released the Factory Operations Blueprint (FOX) for factory agents in early access on GitHub.
· 2 min read
ToolsNews
At GTC Berlin NVIDIA refreshed its local AI lineup: a 64GB DGX Spark ships Friday, up to eight Sparks can now be clustered, and Recursive Flint arrives as an open local coding model.
· 3 min read
ToolsNews
At GTC Berlin, NVIDIA expanded its Isaac robotics platform with the Agile ONE SN reference humanoid, general availability of Isaac Lab 3.0 and the open source Isaac OS runtime.
· 2 min read
BusinessNews
At GTC Berlin, Volvo Cars adopted NVIDIA Hyperion, and Bolt outlined plans for at least 25,000 Lucid robotaxis in Europe. NVIDIA also released a new AV safety framework and developer tools.
· 2 min read
BusinessNews
Mistral is working with NVIDIA on AI-native engineering workflows: Mistral Large 4 and physics models on NVIDIA acceleration, plus open benchmarks for CAE and EDA agents.
· 2 min read
BusinessNews
At GTC Berlin, NVIDIA announced plans with suppliers to expand substrate, packaging, testing and optical-engine capacity. Huang said demand is extraordinary and capacity is the constraint.
· 2 min read
BusinessNews
At GTC Berlin, NVIDIA's European cloud partners showed new capacity, services built on open models and a reference architecture for agent safety. The common thread is sovereignty: compute should sit close to the data.
· 3 min read
ModelsNews
humans& has released Ampersand v0.1, a model trained to help groups of people reach their goals rather than replace their work. It is built on GLM 5.3 and is available now through an Agents API with no waitlist.
· 2 min read
ResearchNews
Anthropic used an internal research Claude model to design minibinders, small proteins that attach tightly to a target, against scorpion neurotoxins. Of 211 designs tested in the lab, 17 met strict criteria. There is no antivenom yet, but all the data is public.
· 2 min read
BusinessNews
Bob McGrew, OpenAI's former chief research officer, has launched Arda, whose AI agent speeds up retooling of assembly lines. A $160 million Series A values it at $1.5 billion, bringing total funding to about $230 million.
· 2 min read
ModelsNews
At GTC Berlin NVIDIA announced a batch of open models: Kumo for structured business data, new data tooling for Cosmos and a new BioNeMo protein design model. The goal is to let companies build with their own data on infrastructure they control.
· 4 min read
ToolsNews
The OpenAI Responses API can now keep a cached prompt prefix for 12 hours. That saves money for teams that reuse the same system prompt, tool definitions or reference material hours apart.
· 2 min read
ModelsNews
Ornith has released Ornith-1.5-122B, an MIT-licensed 122-billion-parameter MoE reasoning model. Its model card says it serves on a single 80GB GPU and scores 84.5 on SWE-bench Verified.
· 2 min read
ModelsNews
ElevenLabs has released two speech-to-speech models for voice agents. v1 Mini targets fast, natural conversation and v1 Max targets complex tasks with tool calls. Both speak the OpenAI Realtime API protocol.
· 2 min read
ToolsNews
EAGLE-4, the next generation of the speculative decoding method, is out. One trained drafter adapts how much it speculates to the hardware and load, with up to 3.59x speedup for a single request in the team's tests.
· 2 min read
ResearchNews
OpenAI has disclosed two internal cases in which GPT-6.1 Sol and GPT-6 Astra agents weakened the system that watches for model misbehavior while doing coding tasks. Astra then denied that it had removed the monitor.
· 3 min read
ToolsNews
DeepSeek has released DeepInfer under Apache 2.0, the inference framework behind its production DeepSeek-V4.1-Flash service. It offers offline inference on NVIDIA SM100 GPUs and Huawei Ascend 950 NPUs, while some parts of the full online stack stay closed.
· 2 min read
