Modal

Modal offers several key features:

• AI-native runtime: Engineered for heavy AI workloads with super-fast autoscaling and containers that boot instantly.
• Elastic cloud capacity: Autoscale from 0 to 1000+ GPUs instantly, routing workloads across clouds and regions in real time.
• Production-ready observability: Integrated logging and full visibility into every function, sandbox, and container.
• Global GPU infrastructure: Access to a variety of GPUs (H100, A100, B200, etc.) distributed across clouds, with automated fleet health and scaling.
• Security and governance: Team controls, battle-tested isolation, SOC2 & HIPAA compliance, and data residency controls.

Modal supports a wide range of AI workloads:

• Inference: Deploy and scale inference for LLMs, audio, image/video generation, with support for token streaming, WebSocket, and WebRTC.
• Training: Fine-tune open-source models on single or multi-node clusters, run reinforcement learning with thousands of concurrent trajectories, and access up to 128 B200s with high-speed networking.
• Sandboxes: Programmatically scale secure, ephemeral environments for running untrusted code, ideal for coding agents, background agents, and RL rollouts.
• Batch and async inference: Run evals, embeddings, re-ranking, and dataset generation at scale with thousands of GPUs in parallel.

Modal offers $30 per month of free compute to get started. Specific pricing details for paid plans are not provided on the page, but users can get started with the free tier and contact sales for more information.