Jake Makler’s Post

4 LLMs in one place - meet Perplexity ! Perplexity AI has recently unveiled two groundbreaking models, pplx-7b-online and pplx-70b-online. These models represent a significant advancement in the realm of Large Language Models (LLMs), addressing key limitations of current LLMs in terms of freshness and accuracy. Key Features: 👉 Up-to-Date Information: Unlike traditional LLMs, these models can access and leverage information from the internet, ensuring responses are not only helpful and factual but also current. 👉 Advanced Base Models: Built upon mistral-7b and llama2-70b, these models are fine-tuned to effectively utilize web snippets for informed responses. 👉 In-House Search Technology: Perplexity AI's own search, indexing, and crawling infrastructure augments the LLMs with relevant and recent information, prioritizing high-quality, non-SEOed sites. Performance Evaluation: 👉 The models were rigorously evaluated against criteria like helpfulness, factuality, and freshness. 👉Human evaluators preferred the PPLX models over others like gpt-3.5 and llama2-70b, particularly for their accurate and up-to-date answers. Accessibility and Pricing: 👉 These models are now publicly available via the pplx-api, marking a significant step in LLM accessibility. 👉 Users can also explore these models through Perplexity Labs, an LLM playground. A new usage-based pricing structure has been introduced, with Pro users receiving monthly credits. Perplexity AI's pplx-7b-online and pplx-70b-online models set a new standard in LLM technology, offering more accurate, timely, and relevant responses. This development marks a significant leap forward in the capabilities of LLMs, particularly in real-world applications where current information is crucial.

OpenAI has just unveiled GPT-4o, the latest version of its artificial intelligence chatbot, and it promises to revolutionise the way we interact with technology. With its ability to respond to audio inputs in as little as 232 milliseconds—comparable to human response times—GPT-4o opens up incredible possibilities, including real-time language translation. This cutting-edge update offers near-instant results across text, vision, and audio, making real-time 'conversations' with the chatbot more seamless and interactive than ever before. The new model is significantly better at understanding visuals and sounds, making it possible to assist visually impaired individuals in navigating cities like London. Moreover, GPT-4o can handle any combination of text, audio, and image inputs and outputs, broadening its applications tremendously. In light of concerns about bias, fairness, and misinformation, OpenAI has subjected GPT-4o to rigorous testing by 70 external experts. This rollout comes amid fierce competition among tech giants to develop advanced AI tools, with Google set to reveal its new AI features shortly. For users, GPT-4o will be available for free, with premium 'Plus' users benefiting from a greater capacity limit for messages. As always, responsible use is key, especially in educational settings where previous versions raised concerns about cheating. This announcement marks another significant milestone for OpenAI, following the phenomenal success of ChatGPT, which became the fastest app to reach 100 million active monthly users just two years ago. https://lnkd.in/ePHguFKf #artificialintelligence #technology #innovation

🚀 Exciting News from OpenAI! 🚀 OpenAI has just unveiled its latest AI model, GPT-4o, and it’s set to revolutionize the way we interact with technology! Why GPT-4o is a Game-Changer: 🎙️ Incredibly Realistic Voice Conversations: Imagine talking to an AI that responds instantly and naturally, allowing you to interrupt and engage just like a real conversation. GPT-4o brings this vision to life! 📸 Seamless Multimodal Interactions: Whether it’s text or images, GPT-4o can handle it all. During a live demo, it even guided a researcher through solving a math equation using its vision and voice capabilities—simply amazing! 🌍 Real-Time Language Translation: Breaking language barriers has never been easier. GPT-4o’s real-time translation abilities make it a versatile tool for global communication. 💬 Engaging and Human-Like Interactions: OpenAI showcased playful and natural dialogues with GPT-4o, demonstrating its ability to make interactions not just functional but enjoyable. Strategic Insights: OpenAI’s CEO Sam Altman highlighted how these advancements make AI interactions feel like scenes straight out of a sci-fi movie. This leap forward positions OpenAI at the forefront of AI innovation, backed by Microsoft’s support. Accessibility for All: Chief Technology Officer Mira Murati announced that GPT-4o will be available for free, with premium users getting extra capacity. Plus, all users now have access to a "browse" feature for up-to-date web information, without relying on ads. As AI continues to evolve, these breakthroughs from OpenAI are setting new standards for interaction, efficiency, and accessibility. We’re excited to see how GPT-4o will transform the AI landscape and enhance user experiences worldwide. We’d love to hear your thoughts! How do you think GPT-4o’s new features will impact the future of AI interactions? Share your insights and let’s start a conversation! Read More: https://lnkd.in/gCgE3Zhx #AI #ArtificialIntelligence #OpenAI #Innovation #Technology #GPT4o #VoiceAI #MultimodalAI #RealTimeTranslation #TechNews

Perplexity AI, released its own AI large language models (LLMs) — pplx-7b-online and pplx-70b-online, named for their parameter sizes, 7 billion and 70 billion respectively. They are fine-tuned and augmented versions of the open-source mistral-7b and llama2-70b models from Mistral and Meta. A distinctive feature of Perplexity's new LLMs is that they are not only accessible for other organizations to build their applications on top of them through Perplexity's API (application programming interface), but they also aim to provide "helpful, factual, and up-to-date information" - something that most other leading LLMs, such as OpenAI's GPT-3.5 and GPT-4 (which power ChatGPT), have difficulty with.

OpenAI's new model GPT-4o allows users to interact using voice, video or text in a single omnipotent system : OpenAI has debuted GPT-4o, an AI model that allows real-time interactions using voice, video or text within a single system for faster responses and smoother transitions between mediums. #AIWorldToday #AIUpdates #ArtificialIntelligence

Google's chatbot AI 'Bard' outperforms GPT-4, claims second place #AIadvancements 🤝 Follow us on Discord 🔜: https://lnkd.in/gt823Zd3 🤝 Follow us on Whatsapp 🔜 https://wapia.in/wabeta _ ❇️ Summary: Large Model Systems Org (LMSYS Org) is a research organization collaborating with several universities to provide large-scale machine learning model datasets, open models, and evaluation tools. They have developed a benchmark platform for large-scale language models and reported that Google's chatbot AI ' Bard with Gemini Pro ' has ranked 2nd, surpassing some models of OpenAI's GPT-4. LMSYS Org is also developing a benchmarking platform called 'Chatbot Arena' that invites human users to chat with AI models and ranks them using the Elo rating. They are looking forward to the release of Gemini Ultra-based Bard in the future. Hashtags: #chatGPT #AIbreakthrough #GoogleBard

Perplexity AI Aims to Revolutionize Web Search with Real-Time LLMs Perplexity AI, founded by ex-Google researchers, aims to challenge Google's web search dominance with its AI chatbot, Perplexity Copilot, and new AI large language models (LLMs) — pplx-7b-online and pplx-70b-online. These LLMs, available via Perplexity's API, boast real-time web search data without knowledge cutoff, outperforming competitors in providing current, accurate, and helpful information. In evaluations against OpenAI's GPT-3.5 and Meta's Llama 2, Perplexity's models excelled in "freshness" and "factuality." The move reflects a broader trend in the AI industry to enhance chatbots with up-to-date knowledge, challenging traditional search engines. Perplexity's API is now open to the general public, with pricing details available through direct inquiry. Aravind Srinivas #PerplexityAI #AI #LLM #WebSearch

OpenAI just dropped a ground-breaking model that takes us one step closer to AGI... Are you ready for a game-changing leap in AI capabilities? OpenAI just dropped a bombshell with their latest announcement of GPT-4o, and let me tell you, it's absolutely mind-blowing! 🤯 1. Seamless Multimodal Integration GPT-4o is a groundbreaking model that can seamlessly understand and generate any combination of text, audio, and visuals in real-time. Say goodbye to the limitations of single-modality AI! 🌉 2. Lightning-Fast Response Times This AI powerhouse can respond to audio inputs in as little as 232 milliseconds, with an average of 320 milliseconds – practically human-like conversation speeds! No more awkward pauses mid-chat. ⚡ 3. Multilingual Mastery While matching GPT-4 Turbo's performance in English and coding, GPT-4o significantly outperforms existing models in understanding and generating content across multiple non-English languages. 🌐 4. Unparalleled Visual and Audio Prowess GPT-4o's vision and audio understanding capabilities are unmatched, setting new benchmarks and leaving other models in the dust. Get ready for a whole new level of AI-powered multimedia experiences! 👁🎵 5. Cost-Effective and Efficient Not only is GPT-4o 2x faster than GPT-4 Turbo, but it also comes at half the price with 5x higher rate limits. Talk about a win-win situation for developers and businesses alike! 💰🚀 OpenAI's GPT-4o is a true game-changer, pushing the boundaries of what AI can do and paving the way for more natural and intuitive human-computer interactions. It seems like we really are getting close to AGI! Let me know your thoughts on this groundbreaking announcement in the comments below. Are you as excited as I am? 🤩

Paid Subscription Customers Can Access OPENAI GPT 4 Chat. #AIConversations Follow us on Discord: https://lnkd.in/gt823Zd3 .. Summary: As a paid subscriber to OPENAI GPT 4, I expected uninterrupted access to the chat service. However, my access was limited, resulting in a shorter usage period. OPENAI GPT 4 is an advanced AI system that engages in natural language conversations with users, offering information and answering questions. Despite the limitations, OPENAI GPT 4 continues to evolve and improve, potentially leading to extended and optimized access for paid subscribers in the future. Overall, while my experience was not ideal, the chat service holds promise for future enhancements. Hashtags: #chatGPT #GPT4ChatSubscribers #PremiumConversations

AI new launches : Old wine , new bottle ? Voice AI dates back to 1900’s with Bell Labs launching a device called “Audrey” which could recognise limited voice commands. ChatGPT, which got 100Mn users in just 2 months is at the vanguard of Gen AI innovation . The new offering by OpenAI’s GPT-4o, is brought to market to highlight its advancements over GPT-4. GPT-4o offers faster and more efficient handling of text, voice, and video inputs, with notable improvements in interpreting images and real-time video. It aims at enhancing accessibility and functionality for a broader audience. GPT-4o's capabilities include live translation, contextual discussions, and a new voice mode for natural, real-time conversations GPT-4o ("o" for "omni"), showcased impressive new features, including a humanlike voice and the ability for varied tasks - e.g. - to tutor children in solving geometry problems. However, the actual release may not entirely deliver many of these advertised features, including the improved voice and real-time video analysis, which were delayed for further refinement. OpenAI also deactivated a previously available voice that resembled actress Scarlett Johansson after legal threats, replacing it with a different voice. The current updates include the ability to upload photos for analysis and quicker, more optomised responses. From a user interaction standpoint, the voice remains machine-loaded. In practical use, while the chatbot excelled at language translations, it struggled with math and physics problems. As per reports, it gave incorrect solutions to some geometry and physics problems but could correct its mistakes with feedback. Despite some improvements, such as correct division calculations, it still failed at basic tasks like counting syllables in words . The new version is free to try on OpenAI’s website and app, with non-paying users having limited access and subscribers getting more usage. OpenAI's iterative approach aims to gather user feedback for continuous improvements. Although there are some enhancements, significant gaps remain especially in reasoning and advanced problem-solving. Ultimately, we must accept that there is a need for real-world testing, continuous human intervention correction, and feedback in refining AI capabilities. Annurag Batra Delshad Irani Prasad Sangameshwaran Raahil Chopra