Brief

AE Studio, a consulting partner for Modal, has developed a workflow for training AI models to prove mathematical theorems using reinforcement learning. They compared two methods: Group Relative Policy Optimization (GRPO) and Evolution Strategies (ES), finding ES to be a promising alternative for this task. The setup leverages Modal's infrastructure for parallel GPU inference and isolated CPU verification, streamlining the research process and accelerating AI-enabled scientific discovery. AI

OpenAI has launched its new Agents SDK, enabling developers to build custom agentic systems for tasks like coding and research. The SDK integrates with Modal's platform, allowing agents to run within isolated sandbox environments, complete with access to resources like GPUs. This integration aims to provide developers with the tools to create powerful internal agentic tools, exemplified by a coding agent capable of parallelizing tasks for challenges like parameter golf. AI

Modal has developed a system to achieve truly serverless GPUs for AI inference, addressing the challenge of rapidly scaling resources to meet variable demand. Their approach involves maintaining cloud buffers of idle GPUs, a custom filesystem for lazy container image serving, and efficient checkpoint/restore mechanisms for both CPU and GPU processes. This engineering effort, developed over five years, reduces AI inference replica scaling time from tens of minutes to mere seconds, aiming to maximize GPU Allocation Utilization. AI

Modal has introduced an autoscaling feature for GPUs designed to support AI research agents. This new capability allows agents to dynamically provision and release compute resources as needed, addressing the challenge of managing unpredictable research workloads and associated costs. The system demonstrated its effectiveness by completing OpenAI's Parameter Golf challenge significantly faster than a traditional workstation, highlighting its efficiency and cost-effectiveness for AI-driven research. AI

Modal has acquired Butter, a company specializing in agent harness engineering. The acquisition brings Butter's founder, Erik Dunteman, and researcher Raymond Tana to Modal's Sandbox team. Their expertise, particularly in creating lightweight sandboxes like bVisor, is expected to enhance Modal Sandboxes into a leading product. AI

Physical Intelligence (Pi) is developing a general-purpose robotic intelligence system using a Visual-Language-Action (VLA) model for real-time control. To enable rapid experimentation and the use of larger models, Pi utilizes Modal's cloud infrastructure for remote inference. They collaborated with Modal to create a specialized QUIC-based transport over UDP, minimizing network latency to approximately 10-15ms and avoiding the potential for TCP-related stalls in the robot's control loop. AI

Modal has released several product updates, including the availability of NVIDIA's RTX Pro 6000 Blackwell GPUs for inference and fine-tuning tasks. The platform also introduced a new Command Palette accessible via keyboard shortcut for easier navigation within the dashboard. Additionally, Modal's Sandbox Filesystem API has entered beta with improved reliability, and the SDK has been updated with enhanced CLI log fetching and new deployment strategies. AI