Nemotron 3 Nano Omni — the first multimodal model in the Nemotron 3 family, now on DeepInfra!

At DeepInfra we host the best open source LLM models. We are always working hard to make our APIs simple and easy to use.
Today we are excited to announce a very easy way to quickly try our models like Llama2 70b and Mistral 7b and compare them to OpenAI's models. You only need to change the API endpoint URL and the model name to quickly see if these models are a good fit for your application.
Here is a quick example of how to use the OpenAI Python client with our models:
import openai
# Point OpenAI client to our endpoint
openai.api_base = "https://api.deepinfra.com/v1/openai"
# Just leave the API key empty. You don't need it to try our models.
openai.api_key = ""
# Your chosen model here
MODEL_DI = "meta-llama/Llama-2-70b-chat-hf"
chat_completion = openai.ChatCompletion.create(
model="meta-llama/Llama-2-70b-chat-hf",
messages=[{"role": "user", "content": "Hello world"}],
stream=True,
)
# print the chat completion
for event in chat_completion:
print(event.choices)
To make it as simple as possible you don't even have to create an account with DeepInfra to
try our models. Just pass empty string as api_key and you are good to go. We rate limit the
unauthenticated requests by IP address.
When you are ready to use our models in production, you can create an account at DeepInfra and get an API key. We offer the best pricing for the llama 2 70b model at just $1 per 1M tokens. If you need any help, just reach out to us on our Discord server.
Langchain improvements: async and streamingStarting from langchain
v0.0.322 you
can make efficient async generation and streaming tokens with deepinfra.
Async generation
The deepinfra wrapper now supports native async calls, so you can expect more
performance (no more t...
What Is Google TurboQuant and What Does It Mean for Open Source Inference? - Deep Infra<p>In late March 2026, Google Research published a paper that got more attention outside of academic circles than most AI research does. TurboQuant, a new compression algorithm for the key-value cache in large language models, landed with enough noise that Cloudflare CEO Matthew Prince called it Google’s DeepSeek moment. The Silicon Valley Pied Piper comparisons […]</p>
Reliable JSON-Only Responses with DeepInfra LLMs<p>When large language models are used inside real applications, their role changes fundamentally. Instead of chatting with users, they become infrastructure components: extracting information, transforming text, driving workflows, or powering APIs. In these scenarios, natural language is no longer the desired output. What applications need is structured data — and very often, that structure is […]</p>
© 2026 Deep Infra. All rights reserved.