OpenAI acquired NextSlide, a startup that turns prompts, notes, documents, and research into editable presentations. The Decoder reported that the acquisition will bring NextSlide's AI-generated presentation capability into ChatGPT.
NextSlide's product focuses on converting four types of source material—prompts, notes, documents, and research—into presentations users can edit.
Why it matters: ChatGPT users working from prompts, notes, documents, or research could gain an editable presentation workflow alongside the chatbot's existing conversational interface.
Meta’s Superintelligence Labs released Muse Glimmer’s 30-billion-parameter weights on Hugging Face under the Apache 2.0 license. Meta says the model can run on a consumer GPU and supports local coding, function calling, local agents, and LLM-as-a-judge evaluation.
The Hugging Face description positions Muse Glimmer as local, agentic, multimodal, and open source. That gives developers a model aimed at agent workflows with a permissive license and consumer-GPU execution.
Meta has not published benchmark results, hardware requirements, or a measured performance comparison in the available material. Developers therefore have no sourced basis yet for judging Muse Glimmer’s cost or capability against another model.
Why it matters: For developers building local coding and agent workflows, Meta offers a 30-billion-parameter model under Apache 2.0, but the missing benchmark and hardware figures leave its practical performance unquantified.
Anthropic will turn on Claude Code's auto mode by default, TechCrunch AI reports. The programming tool will soon require less human oversight.
Why it matters: Claude Code users will get a default programming mode that requires less human oversight once Anthropic enables auto mode by default.
NVIDIA Magpie TTS is presented as a way to build low-latency, multilingual voice agents.
The project highlights open weights and full deployment control for developers building those agents. No performance benchmarks, language count, licensing terms, or pricing details are included.
Why it matters: Magpie TTS gives voice-agent developers a named option for building multilingual systems with open weights and full deployment control, while the available material makes no performance claim.
GeoPT helps AI models understand the basics of physics so they can simulate how objects respond to things like wind and water more efficiently and accurately.
MIT AI News describes the system as a way for AI models to simulate a wider range of real-world scenarios.
Why it matters: AI model developers get a method for expanding the real-world situations their systems can simulate.