DeepInfra

Active

Tagline:AI inference cloud with OpenAI-compatible API and open-source models

Updated:October 22, 2026

Domain Rating:

Social Media:

AI API Interface AI Developer Tools API Available Cloud-based For Developers For Startups For Enterprise Open Source AI Agent Ready Real-time

Open Website

Visit Website

What is DeepInfra?

DeepInfra is an AI inference cloud designed to run open-source machine learning models with an OpenAI-compatible API. It supports LLMs, vision, embeddings, speech, and multimodal architectures with pay-per-token pricing and dedicated GPU deployments on enterprise hardware.

How to use DeepInfra?

1. Sign in to your DeepInfra account and create an API key from the dashboard. 2. Set your environment variable: export DEEPINFRA_API_KEY=<your-key>. 3. Point your OpenAI SDK or HTTP client base URL to https://api.deepinfra.com/v1/openai and make chat completion or inference requests.

DeepInfra Core Features

OpenAI-Compatible API

Drop-in replacement for OpenAI SDKs at https://api.deepinfra.com/v1/openai without code migrations.

Extensive Model Catalog

Host and run 100+ models covering text generation, vision, embeddings, speech recognition, and video.

Pay-Per-Token & Execution Pricing

Billed strictly per token or inference execution time with no minimum commitments, seat fees, or idle GPU costs.

Private Model Deployment & GPU Rental

Deploy custom fine-tuned weights or rent dedicated instances across NVIDIA A100, H100, H200, B200, and B300 GPUs.

High Concurrency & Prompt Caching

Supports 200 default concurrent requests per model alongside prompt caching with retention intervals to reduce latency and cost.

DeepInfra Use Cases

Integrating open-source LLMs into production applications using standard OpenAI client libraries

Building RAG and semantic search pipelines with cost-effective embedding and reranking endpoints

Deploying custom fine-tuned weights and LoRA adapters on dedicated, auto-scaling GPU infrastructure

Transcribing audio and generating speech using high-throughput Whisper and TTS models

DeepInfra Frequently Asked Questions

How do I switch from OpenAI to DeepInfra?

DeepInfra provides an OpenAI-compatible API endpoint at https://api.deepinfra.com/v1/openai. You can configure the standard OpenAI SDK by updating base_url to DeepInfra and passing your DeepInfra API key, requiring no other code changes.

What are the default rate limits on DeepInfra?

Every account receives a default limit of 200 concurrent requests per model. Querying multiple models simultaneously expands your total concurrent limit proportionally.

How does DeepInfra billing work?

DeepInfra operates on a pay-as-you-go model. Language models are billed per token with discounted cached prompt rates, while other models are billed for inference execution time. There are no idle GPU fees or long-term contracts.

Can I deploy custom models or LoRA adapters?

Yes. DeepInfra allows deploying private models and LoRA adapters on dedicated GPU instances with automated scaling.

DeepInfra Product Background & Official Links

DeepInfra Support Email & Customer service contact & Refund contact etc.

DeepInfra Support Email: feedback@deepinfra.com

More Contact, visit the about us page(https://docs.deepinfra.com)

DeepInfra Company Info

DeepInfra Company name: Deep Infra Inc.

DeepInfra Developer Type: company

DeepInfra Company address: DE, US

More about DeepInfra, Please visit the about us page(https://docs.deepinfra.com)

DeepInfra Pricing

DeepInfra Pricing Link: https://deepinfra.com/pricing

DeepInfra Twitter

DeepInfra Twitter Link: https://x.com/DeepInfra

DeepInfra LinkedIn

DeepInfra LinkedIn Link: https://linkedin.com/company/deep-infra

DeepInfra GitHub

DeepInfra GitHub Link: https://github.com/DeepInfra

DeepInfra Pricing & Plans

Pay-As-You-Go

Recurring

From $0.06 / 1M tokens

No upfront cost or subscription fee
Billed strictly per input/output token or execution time
Prompt caching discounts up to 80%
200 default concurrent requests per model
Access to 100+ open-source models

Subscribe & upgrade now

GPU Instances

Recurring

Hourly compute rates

Dedicated NVIDIA A100, H100, H200, B200, and B300 GPUs
Autoscaling and custom container support
Private model and LoRA adapter deployment
Full hardware isolation

Subscribe & upgrade now

Enterprise

contact

Contact Sales

Dedicated clusters and reserved capacity
Custom rate limit increases
Enterprise SLAs and prioritized support
Custom billing terms and invoicing

Subscribe & upgrade now

DeepInfra Tags

API Available Cloud-based For Developers For Startups For Enterprise Open Source AI Agent Ready Real-time Privacy-Focused

Best Suited Occupations for DeepInfra

Software Developers Data Scientists Indie Hackers Product Managers

DeepInfra

Active

Tagline:AI inference cloud with OpenAI-compatible API and open-source models

Updated:October 22, 2026

Domain Rating:

75Domain Rating by Ahrefs

Social Media:

AI API Interface AI Developer Tools API Available Cloud-based For Developers For Startups For Enterprise Open Source AI Agent Ready Real-time

Open Website

Visit Website

What is DeepInfra?

How to use DeepInfra?

DeepInfra Core Features

OpenAI-Compatible API

Drop-in replacement for OpenAI SDKs at https://api.deepinfra.com/v1/openai without code migrations.

Extensive Model Catalog

Host and run 100+ models covering text generation, vision, embeddings, speech recognition, and video.

Pay-Per-Token & Execution Pricing

Billed strictly per token or inference execution time with no minimum commitments, seat fees, or idle GPU costs.

Private Model Deployment & GPU Rental

Deploy custom fine-tuned weights or rent dedicated instances across NVIDIA A100, H100, H200, B200, and B300 GPUs.

High Concurrency & Prompt Caching

Supports 200 default concurrent requests per model alongside prompt caching with retention intervals to reduce latency and cost.

DeepInfra Use Cases

Integrating open-source LLMs into production applications using standard OpenAI client libraries

Building RAG and semantic search pipelines with cost-effective embedding and reranking endpoints

Deploying custom fine-tuned weights and LoRA adapters on dedicated, auto-scaling GPU infrastructure

Transcribing audio and generating speech using high-throughput Whisper and TTS models

DeepInfra Frequently Asked Questions

How do I switch from OpenAI to DeepInfra?

What are the default rate limits on DeepInfra?

Every account receives a default limit of 200 concurrent requests per model. Querying multiple models simultaneously expands your total concurrent limit proportionally.