Launch
AWS Lambda Now Runs GPU Inference with 48 Concurrent Invocations Sharing One Model per GPU
InfoQ · By Steef-Jan Wiggers ·
The Brief
- AWS Lambda now runs Event Functions on EC2 G-series GPU instances through Lambda Managed Instances.
- Up to 48 concurrent invocations can share one model per GPU.
- vLLM and SGLang base images need no handler code, and asynchronous invocations get 90-minute timeouts.
Why it matters
Developers can run GPU-backed model inference on Lambda with shared models and concurrency, without writing a custom handler when using the vLLM or SGLang base images.
Who’s affected
- Teams running GPU inference on AWS Lambda
AWS Lambda now runs Event Functions on EC2 G-series GPU instances through Lambda Managed Instances, with up to 48 concurrent invocations sharing one model per GPU, vLLM and SGLang base images that need no handler code, and 90-minute asynchronous timeouts. Billing is the instance…