The race to build autonomous AI agents has hit a massive bottleneck: data. While frontier models like Claude Code and Codex CLI have demonstrated impressive proficiency in terminal environments, the training strategies and data mixtures behind them have remained closely guarded secrets. This lack of transparency has forced researchers and devs into a costly cycle […]
Liquid AI has released LFM2-24B-A2B, a model optimized for local, low-latency tool dispatch, alongside LocalCowork, an open-source desktop agent application available in their Liquid4All GitHub Cookbook. The release provides a deployable architecture for running enterprise workflows entirely on-device, eliminating API calls and data egress for privacy-sensitive environments. Architecture and Serving Configuration To achieve low-latency execution
This analysis was produced by an AI financial research system. All data is sourced exclusively from publicly available filings, earnings transcripts, government data, and free financial aggregators — no proprietary data, paid research, or institutional tools are used. Every figure cited can be independently verified by the reader at the company’s official Investor Relations website…
Saudi Arabian Oil Company (2222.SR) — AI Equity Research | March 2026 Read More »
Perplexity has released pplx-embed, a collection of multilingual embedding models optimized for large-scale retrieval tasks. These models are designed to handle the noise and complexity of web-scale data, providing a production-ready alternative to proprietary embedding APIs. Architectural Innovations: Bidirectional Attention and Diffusion Most Large Language Models (LLMs) utilize causal, decoder-only architectures. However, for embedding tasks,
In the escalating ‘race of “smaller, faster, cheaper’ AI, Google just dropped a heavy-hitting payload. The tech giant officially unveiled Nano-Banana 2 (technically designated as Gemini 3.1 Flash Image). Google is making a definitive pivot toward the edge: high-fidelity, sub-second image synthesis that stays entirely on your device. The Technical Leap: Efficiency over Scale The
The generative AI race has long been a game of ‘bigger is better.’ But as the industry hits the limits of power consumption and memory bottlenecks, the conversation is shifting from raw parameter counts to architectural efficiency. Liquid AI team is leading this charge with the release of LFM2-24B-A2B, a 24-billion parameter model that redefines
Several years ago, Suzanne Berger was visiting a manufacturing facility in Ohio, talking to workers on the shop floor, when a machinist offered a thought that could serve as her current credo. “Technology takes a step forward—workers take a step forward too,” the employee said. Berger, to explain, is an MIT political scientist who for
A boost for manufacturing Read More »
The Massachusetts Bay Transportation Authority moves hundreds of thousands of people across Greater Boston each day—thanks to a vast system of buses, trains, and ferries that depends on coordination among thousands of employees. In this storied transit system, history runs deep: The Green Line still passes through the country’s oldest subway tunnels, built beneath the
Innovation on the move Read More »
A photogenic green-eyed Russian Blue named Petra might just be the world’s most sequenced cat. Petra was rescued from an animal shelter in Reno, Nevada, by Charlie Lieu, MBA ’05, SM ’05, a data whiz, serial entrepreneur, investor, and cofounder of Darwin’s Ark, a community science nonprofit focused on pet genetics. Since becoming Lieu’s furry
Using big data for good Read More »
Launching from the Lab: Building a Deep-Tech StartupBy Lita Nelsen ’64, SM ’66, SM ’79, former director of the MIT Technology Licensing Office, and Maureen StancikBoyce, SM ’91, SM ’93, PhD ’95, with Sophie Hagerty MIT PRESS, 2026, $35 Empty Vessel: The Story of the Global Economy in One BargeBy Ian Kumekawa, lecturer in historyPENGUIN RANDOM
Recent books from the MIT community Read More »
