DeepInfra raises $107M Series B to scale the inference cloud — read the announcement

The long awaited Llama 2 models are finally here! We are excited to show you how to use them with DeepInfra. These collection of models represent the state of the art in open source language models. They are made available by Meta AI and the license allows you to use them for commercial purposes. So now is the time to build your next AI application with Llama 2 hosted by DeepInfra, and save a ton of money compared to OpenAI's API.
There are 3 different sizes of Llama 2 models as well as chat variants of each size:
Depending on the application you are building, you might want to use a different model. Smaller models are faster and cheaper to run, per token generated. Larger models take longer to run and cost more per token generated, but they are more accurate.
Simply create an account on DeepInfra and get yourself an API Key.
# set the API key as an environment variable
AUTH_TOKEN=<your-api-key>
Each model has a detailed API documentation page that will guide you through the process of using it. For example, here is the API documentation for the llama-2-7b-chat model.
Making an inference request is as easy as making a POST request to our API.
curl -X POST \
-d '{"input": "Who is Bill Clinton?"}' \
-H "Authorization: bearer $AUTH_TOKEN" \
-H 'Content-Type: application/json' \
'https://api.deepinfra.com/v1/inference/meta-llama/Llama-2-7b-chat-hf'
And you will get output like this:
{
"inference_status" : {
"cost" : 0.00454849982634187,
"runtime_ms" : 9097,
"status" : "succeeded"
},
"request_id" : "RKQsJyO5n7ZLif------",
"results" : [
{
"generated_text" : "Who is Bill Clinton?\n\nAnswer: Bill Clinton is an American politician who served as the 42nd President of the United States from 1993 to 2001. He was born on August 19, 1946, in Hope, Arkansas, and grew up in a poor family. Clinton graduated from Georgetown University and received a Rhodes Scholarship to study at Oxford University. He later attended Yale Law School and became a professor of law at the University of Arkansas.\n\nClinton entered politics in the 1970s and served as Attorney General of Arkansas from 1979 to 1981. He was elected Governor of Arkansas in 1982 and served four terms, from 1983 to 1992. In 1992, Clinton was elected President of the United States, defeating incumbent President George H.W. Bush.\n\nDuring his presidency, Clinton implemented several notable policies, including the Don't Ask, Don't Tell Repeal Act, which allowed LGBT individuals to serve openly in the military, and the North American Free"
}
]
}
It is easy to build AI applications with Llama 2 models hosted by DeepInfra.
If you need any help, just reach out to us on our Discord server.
Function Calling in DeepInfra: Extend Your AI with Real-World Logic<p>Modern large language models (LLMs) are incredibly powerful at understanding and generating text, but until recently they were largely static: they could only respond based on patterns in their training data. Function calling changes that. It lets language models interact with external logic — your own code, APIs, utilities, or business systems — while still […]</p>
Step 3.5 Flash API Benchmarks: Latency, Throughput & Cost<p>About Step 3.5 Flash Step 3.5 Flash is an open-weights reasoning model released in February 2026 by StepFun. It leverages a sparse Mixture of Experts (MoE) architecture with 196 billion total parameters and only 11 billion active parameters per token during inference — delivering state-of-the-art performance at a fraction of the cost of dense models. […]</p>
Step 3.7 Flash is Live on DeepInfra: An Agentic, Multimodal Model Built for ProductionStepFun's Step 3.7 Flash is now live on DeepInfra. It's a 198B-parameter sparse MoE vision-language model with just ~11B active parameters per token, a 256K context window, and three selectable reasoning levels—purpose-built for high-throughput agentic workflows that combine perception, search, and reasoning.© 2026 DeepInfra. All rights reserved.