1. DeepMind alumni startup Inherent says its AI teammate beat larger frontier models at research replication
요약: Inherent, a London lab founded by Google DeepMind alumni, says its Faraday AI agent outperformed larger systems from Anthropic and OpenAI on independently reproducing published scientific findings. The company highlights reinforcement learning, research taste, and smaller-model efficiency, positioning Faraday as a step toward AI scientist agents rather than a general coding assistant.
2. Who’s behind the new ‘stealth model’ Ox Alpha?
URL: https://techcrunch.com/2026/08/23/whos-behind-the-new-stealth-model-ox-alpha/
요약: TechCrunch reports that a mysterious free reasoning model, Ox Alpha, appeared on OpenRouter on August 21 and quickly triggered speculation about its origin. The model is described as targeting coding, sustained agentic work, and production workloads, but its provider remains anonymous during preview, underscoring how opaque model provenance is becoming in routing marketplaces.
3. Up to 3.2x Faster Inference with LFM2.5-DSpark
URL: https://huggingface.co/blog/LiquidAI/lfm25-dspark
요약: Liquid AI published LFM2.5-DSpark on Hugging Face, reporting up to 3.2x faster inference for LFM2.5 through DSpark optimization. The release is relevant for teams trying to reduce serving latency and cost for edge or production deployments while preserving model quality.
Sources: