Inception Labs CEO on diffusion LLMs 10x faster than autoregressive models

MAI-Transcribe-2-Streaming returns partial transcripts in just over 100ms across 60 languages, ranking No. 1 for both final and partial transcript accuracy on Artificial Analysis. Introductory pricing is $0.54 per hour of audio through the end of the year; Microsoft also announced MAI-Voice-2.1 and MAI-Voice-2.1-Flash covering 23 languages and 26 locales.

The new hand has 13 degrees of freedom and is directly actuated, built for high-fidelity simulation to enable sim-to-real reinforcement learning. Boston Dynamics says the previous three-fingered hand was never intended for mass production, and that human-like designs sacrifice reliability and manufacturability.

Under an equal time budget and the same frontier LLM backbone, open-source state-of-the-art MLE harnesses showed no advantage over a single session of a minimal-harness coding agent with read, write, and bash primitives. Apple researchers conclude the backbone is the primary driver and hand-crafted harness layers are redundant for current MLE benchmarks.

PwC's 2027 Global Digital Trust Insights survey of nearly 4,000 business and tech leaders across 70+ countries found 55% cite reliability and maturity failings as the reason they won't allow fully autonomous AI cyber defense. Just 21% are implementing quantum-resistant security measures.

The Odd Lots podcast examines whether AI-focused companies maintain resilience against Federal Reserve interest rate adjustments. The discussion explores if these firms offer a stable investment hedge during periods of broader market volatility.

Photon's seed round follows 40,000+ developers signed up and 10x revenue growth in four months. Its toolkit — unified API, channel framework, CLI, observability suite — lets agents run over iMessage, WhatsApp, Telegram, SMS/RCS, email, and voice.