Blogs›Replicate›GPU Hardware Support and Migration
Replicate has introduced support for NVIDIA H100 GPUs, offering significant performance improvements. Additionally, multi-GPU configurations (2x, 4x, 8x) for A100 and L40S GPUs are now available for regular models and training runs, expanding beyond deployments. Pricing details for these new hardware options have been released, and users can now create models and update deployments to utilize H100s and the expanded multi-GPU setups via the web interface and API.
5/16/2025
This post announces the availability of NVIDIA H100 GPUs and expanded multi-GPU configurations (2x, 4x, 8x) for A100 and L40S GPUs for general model use and training. It details pricing for these new options and provides API examples for creating models and listing available hardware, including specific SKU names for H100 and multi-GPU setups.
11/15/2024
This post announces the addition of NVIDIA L40S GPUs to Replicate's supported hardware, highlighting a ~40% performance improvement over A40 GPUs. It details a phased migration plan for existing models and deployments from A40 to L40S, including specific dates for public and private resources. The post also analyzes the cost implications of the new GPUs, explaining how pricing will be affected for different usage scenarios (private vs. public models/training) and provides guidance on adjusting deployment configurations to optimize costs based on the performance gains.
6/12/2024
Announces the upcoming addition of NVIDIA H100 GPUs to Replicate's supported hardware, promising enhanced performance. It also mentions early access opportunities via email.