Jensen Huang walked onto the GTC stage and said something that did not sound like a chip announcement. He called Vera Rubin "the greatest infrastructure buildout in history." That is a bold claim even for NVIDIA. But when you look at what Vera Rubin actually is the ambition makes more sense. This is not a faster GPU. It is seven chips designed to work together as one supercomputer, built specifically for a world where AI does not just answer questions but plans, executes, and runs continuously without stopping. Every GPU you have used until now was designed for training massive models or answering queries fast. Neither of those is the same as running an agent that plans, executes tools, checks its own work and keeps going for hours. Current infrastructure was simply never designed for that workload. Vera Rubin is NVIDIA's answer to that problem.

Most text-to-speech tools fall into two camps. The ones that sound good need serious hardware. The ones that run on anything sound robotic. MOSS-TTS-Nano is trying to be neither. It's a 100 million parameter model that runs on a regular CPU and it actually sounds good. Good enough that the team behind it built an entire family of speech models around the same core technology, one of which has gone head to head with Gemini 2.5 Pro and ElevenLabs and come out ahead on speaker similarity. It just dropped on April 10th and it's the newest addition to the MOSS-TTS family, a collection of five open source speech models from MOSI.AI and the OpenMOSS team. The family doesn't just cover lightweight local deployment. One of its models MOSS-TTSD outperforms Gemini 2.5 Pro and ElevenLabs on speaker similarity in benchmarks. Another generates voices purely from text descriptions with no reference audio needed. And one is built specifically for real-time voice agents with a 180ms first-byte latency. Nano is the entry point. The family is the story.

daVinci-MagiHuman processes text, video and audio inside a single unified transformer simultaneously. No separate models, no post processing alignment. The lip sync and facial dynamics are not corrected after generation. They are generated correctly from the start because all three streams are being denoised together.

IBM just released Granite 4.1, a family of open source language models built specifically for enterprise use. Three sizes, Apache 2.0 licensed and trained on 15 trillion tokens with a level of pipeline obsession that's worth understanding. But there's one result in the benchmarks I keep coming back to. The 8B model. Dense architecture, no MoE tricks, no extended reasoning chains. It matches or beats Granite 4.0-H-Small across basically every benchmark they ran. That older model has 32B parameters with 9B active. This one has 8 billion. Full stop. That result is either very impressive or it means the old model was underbuilt. Probably both. Here's how they built it, what the numbers actually say, and whether any of it matters for your use case.

Mistral has been shipping specialized models for a while now. One for coding. One for reasoning. One for chat. Each one doing its thing separately and requiring a different deployment decision. Medium 3.5 ends that confusion. One 128B dense model, one set of weights, handling instruction following, reasoning, and coding together. Mistral didn't just release a new model, they retired three existing ones to make room for it. Devstral 2, Magistral and even Medium 3.1 is gone. Medium 3.5 is what replaced all of them. That's either a sign of real confidence or a very expensive consolidation bet. Looking at the benchmarks, it's starting to look like the former.

Nobody expected a 1.15 GB model to score competitively against full precision 8B models. That is not how this usually goes. PrismML released Bonsai 8B last month and the headline number is almost absurd. The whole model, weights and all, fits in 1.15 GB. For context, the standard FP16 version of a comparable 8B model sits at around 16 GB. Bonsai beats or matches several of them on benchmarks while being 14 times smaller. It runs on a phone. There is literally an iPhone build. I want to be clear that these numbers come from PrismML's own evaluations, not independent third party testing. But even with that caveat, this is worth paying attention to.

Voxtral TTS supports nine languages: English, French, German, Spanish, Dutch, Portuguese, Italian, Hindi, and Arabic. That by itself isn’t unusual anymore. A lot of models claim multilingual support. The interesting part is how it handles switching between them. Mistral says it can move between languages mid-sentence without changing the speaker’s voice. So you don’t get that awkward reset where the tone or identity shifts when the language changes. If that holds up, it’s actually useful for real scenarios like think support calls where people naturally switch languages, or content that mixes languages without warning.