DeepInfra has announced an easy approach to test their open-source language models, such as Llama2 70b and Mistral 7b, and compare them with OpenAI's offerings by simply changing the API endpoint URL and model name without requiring an API key or account creation. They provide a sample Python code snippet for users to try their models, emphasizing ease of access with rate limits on unauthenticated requests based on IP address. For production use, DeepInfra offers affordable pricing at $1 per million tokens for the Llama2 70b model and encourages potential users to contact them for further assistance or sales inquiries. Their infrastructure supports scalable model hosting with enterprise-grade uptime and competitive rates, and users can reach out for support through their Discord server.