Sequoia
·
2 months ago
Sequoia Capital hosted its fourth annual AI Ascent conference on April 20, 2026 in San Francisco with over 150 AI founders and researchers including leaders from Anthropic, Waymo, and Nvidia. The event featured discussions on agents as a major focus area, alongside topics including robotics, data centers in space, and neural network science. Participants emphasized building customer-focused AI products that exploit the gap between current model capabilities and enterprise deployment.
OpenAI Blog
·
2 months ago
OpenAI has implemented security measures for Codex including sandboxing, approval systems, network policies, and telemetry designed to enable safe deployment of coding agents. The system uses agent-native telemetry to monitor behavior and network policies to restrict unauthorized connections. These controls allow organizations to adopt coding agents while maintaining compliance and reducing risks from automated code execution.
BAIR
·
2 months ago
Adaptive Parallel Reasoning (APR) is a paradigm where language models dynamically decide when to parallelize reasoning tasks, how many threads to spawn, and how to coordinate them based on problem complexity. Unlike fixed parallelization approaches such as tree search or best-of-N sampling, APR models learn through reinforcement learning to avoid redundant computation and choose not to parallelize on simple problems. The shift enables more efficient inference by matching parallelization level to task difficulty, reducing latency and context window overflow while maintaining reasoning quality.
Allen Institute (AI2)
·
2 months ago
Researchers released EMO, a mixture-of-experts language model with 128 total experts and 1 billion active parameters trained on 1 trillion tokens, designed so that modular structure emerges naturally from data without predefined domains. The model can use just 12.5% of its experts for specific tasks while retaining 97% of full-model performance, whereas standard MoE models degrade sharply when experts are pruned. EMO achieves this by restricting tokens within the same document to activate from a shared expert pool during training, causing experts to organize around semantic domains like health and politics rather than surface features like prepositions.
Together AI
·
2 months ago
Together AI's Dedicated Container Inference service, combined with the Goose agent and a new dedicated containers skill, enables developers to deploy any HuggingFace model with a single sentence prompt. A developer deployed Netflix's void-model video object removal model from release day to production in one session using three command-line steps. This eliminates the typical one to two day lag between discovering a new model and getting it running in a usable environment.
