DeepInfra

What it is

Cost-effective, scalable inference infrastructure for machine learning models with 100+ models, low pay-as-you-go pricing, and US-based data centers.

Quick answers

What is DeepInfra's data retention policy?

DeepInfra has a zero retention policy where inputs, outputs, and user data stay private. The company is SOC 2 and ISO 27001 certified and follows best practices in information security and privacy.

What pricing models does DeepInfra offer?

DeepInfra offers different pricing models depending on the model used. Language models are charged per token (input and output), most other models are billed for inference execution time, and GPU rental is billed per instance-hour. There are no long-term contracts or upfront costs.

What infrastructure does DeepInfra use?

DeepInfra runs on its own cutting-edge inference-optimized infrastructure in secure US-based data centers. The company owns and operates its hardware rather than relying on cloud providers, which gives better control over cost, performance, and reliability.

When was DeepInfra founded?

DeepInfra was founded in September 2022. The team brings deep experience building backend infrastructure at massive scale, previously having built and operated infrastructure for imo messenger, a platform with over 1 billion Play Store downloads.

What was DeepInfra's Series B funding?

DeepInfra raised $107M in Series B funding to scale the inference cloud. The round was co-led by 500 Global and Georges Harik, with participation from A.Capital Ventures, Crescent Cove, Felicis, NVIDIA, Peak6, Samsung Next, Supermicro, and Upper90.