OpenAI says its August 28-trained AI model solves the Navier–Stokes Millennium Prize problem among more than 100 open math questions, but NYU mathematician Tristan Buckmaster alleges the company front-ran his unpublished work ahead of a Tuesday GitHub release.
Mistral opens a public-preview API for its 1-trillion-parameter multimodal Mistral Large 4, plans to release open weights by October 2026, and is red-teaming a less-moderated cyber-capable version with vetted partners and state authorities.
Mistral AI launches its 1-trillion-parameter multimodal Mistral Large 4, initially restricting access to a public guardrail endpoint before planning to release the model’s weights in three weeks following safety testing.
Google launches EmbeddingGemma 2, a 740-million-parameter Apache 2.0 model that embeds text, code, images, video and audio in one on-device space, supporting 8K-token context and using as little as 191MB of active RAM on a Pixel 11 Pro when quantized.
Aleph Alpha releases Kolibri, a sovereign German-English open-weight AI model for mission-critical government and industry use, combining 78 billion total parameters with roughly 3 billion active per token and training data screened against a 4.5-million-URL blocklist.
Reflection unveils Beam, a 501-billion-parameter Mixture-of-Experts reasoning model with 23 billion active parameters, and says it will release Apache 2.0-licensed weights and developer tools later this month.
OpenAI launches opt-in text watermarking for select API models worldwide and says it will add invisible watermarks to eligible ChatGPT and Codex outputs in the EU within weeks, while restricting its detector to approved researchers because synonym edits can sharply reduce detection accuracy.
Microsoft unveils MAI-Transcribe-2-Streaming, its first real-time transcription model for 60 languages, saying it produces text within 100 milliseconds for live dictation and subtitles, alongside new MAI-Voice-2.1 text-to-speech models covering 23 languages and 26 locales.
Strands launches the open-source Decider 2B model for low-latency local AI-agent decisions, reporting roughly 115 ms median latency on an Nvidia RTX 3090 and a top JevBench ranking among strictly 2B-parameter models.
arXiv imposes a universal two-paper-per-month submission limit starting October 1, 2026, responding to a surge in AI-written, fragmented manuscripts that helped drive 40,363 submissions and nearly 9,000 support tickets in September.