Nov 28, 2023

on November 28, 2023 at 10:00 am Inflection AI has introduced a new iteration of its foundational large language model , Inflection-2, calling it “the best model in the world for its compute class and the second most capable LLM in the world today.” Inflection-2 will be available through the startup as an API and will fuel Inflection’s generative AI chatbot Pi , Inflection-2 Flex Inflection-2 was trained on 5,000 Nvidia H100 GPUs and, according to Inflection, has demonstrated superior performance in benchmarks compared to its predecessor and other leading models like Google’s PaLM 2 Large. The startup claims the model shines particularly in areas such as factual knowledge, stylistic control, and reasoning, outperforming on several standard generative AI benchmark tests. The model’s efficient design and transition to H100 GPUs have led to cost reductions and increased speed despite its larger size compared to Inflection-1. Inflection’s new LLM still reflects the company’s interest in creating an emotionally intelligent AI conversationalist that can replicate empathy and human emotion. Inflection’s online and mobile chatbot Pi, short for Personal Intelligence, will be augmented with Inflection-2 after it passes some extra safety and utility tests. You can see in the chart below how Inflection-2 did against Inflection-1 and Google’s PaLM 2 model. PaLM 2 was often the model comparison chosen by Inflection, which Voicebot founder Bret Kinsella pointed out might hint as to how the startup is pitching its work in our Synthedia newsletter . “The choice to focus on the comparison with PaLM 2 suggests that Inflection is focused on gaining a position as a reasonable OpenAI alternative,” Kinsella wrote. “If you think PaLM 2 is the key alternative right now and it is a safe choice, given its backing by Google, then showing that comparison makes sense. Granted, the more prominent alternative to OpenAI’s LLMs is Antrhopic, but it is only mentioned for the MMLU benchmark.” Inflection Ascendant “Our mission at Inflection is to create a personal AI for everyone. Just a few months ago, we announced Inflection-1 — a best-in-class language model that currently powers Pi. Our new model, Inflection-2, is substantially more capable than Inflection-1, demonstrating much improved factual knowledge, better stylistic control, and dramatically improved reasoning,” Inflection wrote in a blog post. “This is a big milestone on our path towards building a personal AI for everyone, and we’re excited about the new capabilities that Inflection-2 will enable in Pi. As our scaling journey continues, we are already looking forward to training even larger models on the full capacity of our 22,000 GPU cluster. Stay tuned!” Inflection has plenty of cash to continue its projects after raising $1.3 billion in a funding round led by giants of tech like Nvidia, Microsoft, Bill Gates, and Google co-founder Eric Schmidt. The startup has ambitions to build the largest AI cluster in the world, as mentioned in the announcement above. At 22,000 Nvidia H100 Tensor Core GPUs, it will provide 22 exaFLOPS to run the most precise version of the Inflection-1 model. The completed project will not only outmatch any existing AI cluster; there’s only one computing cluster of any kind that will be bigger.