Muse Spark Open is an 816B mixture-of-experts model with 42B active parameters and a 1M token context window, scoring 39 on the Artificial Analysis Intelligence Index — the #1 US open-weights model. Meta trained it on NVIDIA GB300 NVL72 for agentic coding, and it adds audio and video perception alongside image and document understanding.
Carbon-X mid-trains a 14B model on 300B tokens of DNA and text, beating every open language model tested on five DNA benchmarks and landing within a point of GPT-6.1 Sol at ~1,000x lower inference cost. It runs locally on a laptop and was built with Zhongguancun Academy and University of Naples Federico II.
Reservoir Media filed copyright infringement and DMCA suits against Suno and Anthropic in U.S. court, seeking nearly $340 million in combined damages, with roughly $200 million of that aimed at Suno. Attorney Richard S. Busch represents Reservoir in both actions.
Replit's Monitoring Agent reads production logs and metrics, investigates alerts, validates a fix, and can auto-deploy it to production, then emails a summary of what changed. It builds on App Monitoring, launched earlier this year, which only flagged outages.
DeepInfer, the stack behind DeepSeek-V4.1-Flash, ships under Apache 2.0 and supports Nvidia SM100 (Blackwell) GPUs and Huawei Ascend 950 NPUs. DeepSeek says one cluster serves 155 trillion tokens per day; the release covers offline inference only, without its KV cache, P/D disaggregation or load balancing.
The round is led by Coatue and UAE investors and would make OpenAI the most valuable private company ever, topping its $852B post-money mark from March. CNBC reports the UAE funding could come with a deal for more OpenAI computing capacity there.