Latest stories
The innovative Multi-Token Residual Prediction framework boosts language model efficiency, enhancing throughput and accuracy across various decoding regimes.
Explore the economic implications of using serverless versus reserved GPUs, including cost models, demand prediction challenges, and practical benefits for AI d...
Modal is essentially a computer capable of running complex programs while providing an efficient architecture for processing and data management.
Thinking Machines' Inkling, a multimodal transformer model, integrates text, audio, and visual inputs, prioritizing efficiency and speed in processing.
Modal has reengineered its sandbox platform to support rapid scaling, achieving the creation of one million concurrent sandboxes in under a minute.
Moonshot's Kimi K3, a 2.8 trillion parameter model, excels in multimodal tasks with remarkable speed and efficiency, setting a new benchmark in open AI models.
Modal's new I/O plane significantly reduces latencies for serverless functions by offering regional routing options, improving overall speed and efficiency.
The Qwen 3.8-2.4T-A95B model is now available on Modal, offering notable advancements in coding, research, and tool-based tasks.
Botika's innovative approach automates the visual production pipeline for fashion brands, utilizing a custom-built full-stack AI on Modal for enhanced efficienc...
Modal announces its new London office, marking a significant expansion in Europe aimed at serving the growing AI market.
Modal introduces support for new AI models, budget controls, improved dashboards, and more to enhance user experience and resource management.
Discover essential insights for software engineers to optimize coding agents and enhance inference performance in the evolving technology landscape.
Explore the growing demand for efficient language models that prioritize functionality over advanced intelligence, enhancing cost-performance ratios.
Modal Clusters offer organizations a scalable solution for multi-node compute, enhancing AI model training and inference with efficient resource management.
TypeSafe AI’s Jev model, focused on rapid decision-making, scales to over one trillion tokens in just three days using Modal’s serverless infrastructure.
Endpoint Candidates enhance inference evaluation by allowing users to test deployments against production traffic, optimizing for latency and cost.
Research explores the use of mixed-precision, mixed-accelerator samplers in reinforcement learning, aiming for improved performance and cost-efficiency.
Amazon introduces new shopping features for Prime Video that enhance product discovery while viewing, expanding on previous capabilities.
Amazon introduces a new lineup of Alexa tablets featuring personalized AI, improved processing power, and full Google Play Store access, broadening user experie...
Cisco Talos has introduced CAIRN, an open-source framework for identifying AI-integrated malware. The tool reveals new insights into autonomous hacking tools le...