Meta, which has done notable work in the artificial intelligence industry lately, introduced a new generative artificial intelligence model at an event it hosted.Lama 3.3 70BLlama 3.3 70B, the latest member of the company’s “Llama” model family, has remarkable features.

According to the statement of Ahmad Al-Dahle, one of Meta’s senior managers, the new model is compatible with the “Llama 3.1 405B”, the largest model in the Llama family. offers the same performance. But there is more; new model, lower costs of will succeed in attracting the attention of companies.

Better than competitors like Google and OpenAI!

Llama 3.3 70B, which only appears as a text-oriented artificial intelligence model, is better than models such as Google Gemini 1.5 Pro, OpenAI GPT-4o and Amazon Nova Pro in the tests performed. shows higher performance. At least the chart shared by the Meta official says this is the case. However, an unnamed Meta spokesperson told TechCrunch that the new model will focus on general knowledge, math, app usage, and following instructions. needs to be improved he declared.

According to Meta’s statement, Llama 3.3 is suitable for 1 million entries in 70B. $0.1 affordable. The cost of 1 million outputs is $0.4. The price Meta wants for Llama 3.1 405B is $1 for 1 million inputs and $1 for 1 million outputs. $1.8. This situation will produce businesses in terms of savings It seems like it would be important. Because these prices go up to $10 on the OpenAI and Google front.