The rollout of the advanced Voice Mode was delayed due to the need for further content filtering improvements. Now, OpenAI CEO Sam Altman has confirmed that the alpha rollout is happening next week.
Llm RSS
Microsoft announced the availability of Phi-3-mini and Phi-3-medium small language models for fine-tuning on Azure, allowing developers to customize these models for various use cases.
Gemini chat assistant has replaced the Gemini 1.0 Pro model with the faster and more efficient Gemini 1.5 Flash model. The Gemini 1.5 Flash model supports a 4x larger context window (32k tokens).
Reddit has blocked search engines, including Bing, from crawling its site by updating the robots.txt file. Also, it claims that this change has nothing to do with its recent partnership with Google.
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.
Meta has announced the release of its Llama 3.1 family of AI models, which includes three models: 8B, 70B, and 405B. The new models come with improved context length and much-improved performance.
Leaked benchmarks suggest that Meta's upcoming Llama 3.1 405B model could outperform OpenAI's GPT-4o on several key AI benchmarks, marking a potential milestone for open-source AI.
Apple recently introduced its open-source DCLM-7B model, showcasing the potential of data curation in enhancing model performance. However, the DCLM-7B performs poorly against Microsoft's Phi-3.
In response to the industry's trend of releasing cost-effective AI models, OpenAI announced GPT-4o mini. This new model is cheaper and faster than its predecessors, replacing GPT-3.5 Turbo.
Fujitsu and Cohere have agreed to a strategic partnership, which will see them work together to deliver large language models for businesses and enterprises with a focus on privacy and security.
A team of Microsoft researchers have published a paper that details their work on SpreadsheetLLM, an AL large language model designed specifically to help users with complex spreadsheets.
Chinese search engine giant Baidu has launched its new LLM called Ernie 4.0 Turbo. It's an upgraded version of the Ernie 4.0 model, which has existed since last year and was meant to rival GPT-4.
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.
Nvidia could soon overtake Apple as the second-most valuable company in the world, only behind Microsoft. Nvidia has been doing very well for itself in recent years because it powers ChatGPT.
Not long after OpenAI showcased GPT-4o, Google uploaded a demo video showing a prototype version of Gemini analyzing a camera feed in real time. It's expected to talk more about AI at Google I/O.
Google Gemini has some limitations, one of which is the inability to identify songs or work with third-party music apps. But as per a new tip, Gemini could soon let you choose your music service.
After AI companies hit the wall to gather high-quality data to train their AI models, companies like OpenAI are alleged to have transcribed a million hours of YouTube videos to train GPT-4.
Opera will allow users using Opera One to run LLMs locally on their computers. The browser supports more than 150 models from 50 families including LLMs from Meta, Google, and more.
After planning to bake Google's Gemini features on the iPhone, Apple is also in talks with China's Baidu to bring similar features to iPhones in China amid stringent AI compliance requirements.
The Brave browser's in-built AI tool, Leo, is rolling out to Android users in the coming days. It helps users to summarize and translate pages and more. You can use three popular LLMs with it.
Nvidia, Hugging Face, and ServiceNow have released a new set of LLMs known as StarCoder2. Trained on 619 programming languages, it aims to help programmers writer and understand code efficiently.
Yelp's restaurant reviewing platform is continuing to be updated, with new features being added that are powered by AI such as business summaries and a new home feed which shows relevant photos.
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 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.
Microsoft has announced that it will be introducing four new AI text-to-speech voices optimised for conversations which can be used for chatbots, voice assistants or conversational agents.
Google reportedly is trying to launch its next-gen "Gemini" AI this fall. It will reportedly combine text prompts and image generation, trying to compete with OpenAI and Microsoft.
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.
Arjaldo Karaj, the Director of Product at Brave Search. talks to us about its new feature that offers an independent alternative to searching for images and videos on the internet.
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.
The Brave Search API is now live. The service is free for companies and developers for limited use but also offers several paid tiers for those who have more demanding requirements.
Google has announced a new Large Language Model called PaLM 2 that improves upon its predecessor and is now 25 Google services including specialized services like Med-PaLM 2 and Sec-PaLM.
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.
This week's Microsoft Weekly digest is chock-full of news items related to controversies related to AI-powered products, the arrival of free USBs for Windows Insiders, and multiple app updates.
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.
Microsoft latest endeavor into the AI-powered chatbot space has proven that the company is still grappling with the inherent challenges of the technology, such as curation, biasness, and accuracy.
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.
Jailbreak hacker @ronsoros found stolen, open-source code in voice.ai, a real-time voice synthesizer, but instead of complying with the opensource license, they are taking measures to shut him down.