Modal

Serverless GPU cloud for ML inference, agent sandboxes, and batch jobs.

by Modal · ML Platforms / Inference

Executive Summary

Serverless GPU cloud for ML inference, agent sandboxes, and batch jobs.

Use Cases

  • ML inference
  • LLM inference
  • Agent sandboxes
  • Batch processing
  • ML training

Features

Visibility

  • Billing and Usage Monitoring: Monitor compute usage and costs through a dedicated dashboard.
  • OpenTelemetry Integration: Export Modal logs to any OpenTelemetry-compatible observability provider.

Intelligence

  • Instant Autoscaling: Automatically scales GPU resources from 0 to 1000+ based on demand, with no commitments.
  • Sub-second Cold Starts: Rapid deployment and execution of ML workloads with minimal startup latency.
  • Multi-cloud Orchestration: Routes workloads across clouds and regions in real time to optimize performance and availability.

Support

  • Academic Program Credits: Unlock up to $10k of credits for academic projects to supercharge research.

Technical Specifications

Architecture
Serverless GPU cloud platform that autoscales from 0 to 1000+ GPUs, routes workloads across clouds and regions in real time, and offers sub-second cold starts for ML workloads.
Deployment
SaaS
API Available
Yes

Infrastructure

  • Multi-cloud (AWS, GCP, Azure)

Integrations

  • OpenTelemetry

Security & Compliance

Certifications: SOC 2 Type II, HIPAA (Enterprise plans with BAA)

Pricing

Model
Pay-as-you-go (usage-based)
Starting Price
Usage-based, see pricing page for rates
Target Customer
SMB,Mid-Market,Enterprise
Contract Type
Usage-based, billed per second/minute
Free Trial
Yes (no credit card required)

About Modal

Modal is a serverless compute platform that provides high-performance AI infrastructure. It abstracts away the complexities of GPU infrastructure management, making it easy for developers to run compute-intensive workloads like ML inference, fine-tuning, and AI agent sandboxes.

Founded: 2021 · Headquarters: New York City, United States · Employees: 51-200 · Private