Mistral AI, a French startup, has released Large 2 with 123 billion parameters, a new flagship model with improved capabilities in code generation, mathematics, reasoning, and multilingual support.
Large language model RSS
Amazon Bedrock has made the new Anthropic Claude 3.5 Sonnet model available for developers. The new model which was just released scores better in benchmarks than rival such as GPT-4o.
Chinese regulators have approved over 40 AI models for public use. The process started around six months back, and the first batch of AI models got the green signal in August 2023.
Microsoft released the official Copilot app for iOS. The users of iPhones and iPads can install "your everyday AI companion" through the App Store. The tool offers free access to OpenAI's GPT-4.
Microsoft Research has released Orca 2, a new AI language model in two sizes. Both can match the performance of LLMs like Llama 2 in certain ways, even though they are much smaller.
Spotify is partnering with Google to use its LLMs to improve podcast and audiobook recommendations to users. This could help Spotify boost the amount of time users spend using the app.
Anthropic has launched its Claude 2 LLM for people in 95 countries. It was only available in the US and the UK until Tuesday. Claude 2 can process very large documents and answer questions about them.
Anthropic has announced the availability of Claude Instant 1.2 through an API. Interestingly, benchmarks show it to be the safest of all the models so far with fewer hallucinations.
OpenAI is currently researching process supervision, a method for reducing the number of hallucinations produced by LLMs. It has had success with maths datasets but needs to be tested generally.
Google is developing a generative AI chatbot for Wendy's that will take drive thru orders from a single company-owned restaurant in June. It's likely it will expand to more locations in the future.
It might take a lot of money to run the popular AI bot ChatGPT. A firm estimates that the number could be as high as $700,000 per day, most of which goes into hardware infrastructure used to run it.
Alphabet's investment firm, CapitalG, has provided AI market intel firm, AlphaSense, with $100 million in funding. Part of the money will be used to integrate an LLM to leverage AlphaSense's data.
Brave Search has a new feature now called Summarizer which takes information from webpages and presents relevant information coherently using a large language model, similar to what ChatGPT uses.
Meta is preparing to launch LLaMA for researchers to assist in their work, according to Mark Zuckerberg. The large language model will also use an open model, accessible by researchers.
After companies like Microsoft, Google, and Baidu announced their take on the popular ChatGPT conversational chatbot, e-commerce giant Alibaba has announced its plans to develop their own AI tech.
