
Introducing Mistral Large: Mistral’s most advanced Multilingual Reasoning LLM with Azure Integration

H2O AI’s Danube – A Super-Tiny LLM for Mobile Applications
Other AI News
-
Alibaba’s EMO AI: Revolutionizing Video Generation with Lifelike Talking and Singing Portraits
Researchers at Alibaba’s Institute for Intelligent Computing have unveiled a groundbreaking artificial intelligence system named “EMO” (Emote Portrait Alive), capable of animating still portrait photos to produce videos where the subject appears to talk or sing with lifelike realism. This innovative system leverages a direct audio-to-video synthesis approach, avoiding traditional methods that use 3D models or facial landmarks. By training on over 250 hours of diverse talking head videos, EMO can generate expressive facial movements and head poses that align closely with the audio’s nuances, marking a significant leap forward in the realm of audio-driven video generation.
EMO distinguishes itself by employing a diffusion model, a type of AI that excels in creating realistic synthetic images, to transform audio waveforms directly into video frames. This method captures the subtle motions and unique characteristics of natural speech more accurately than previous technologies. The system not only produces convincing conversational videos but also excels at creating singing videos with synchronized mouth shapes and facial expressions. Despite its potential for creating personalized video content from mere photos and audio clips, the technology raises ethical concerns regarding impersonation and misinformation. The team behind EMO is exploring ways to detect synthetic videos to mitigate these issues.
-
Figure’s Breakthrough: $675M Funding and OpenAI Partnership Propel Humanoid Robotics Forward
Figure, a robotics startup based in Sunnyvale, California, has made headlines with its recent $675 million funding round, bringing its valuation to an impressive $2.6 billion. This investment, contributed by tech giants such as Microsoft, OpenAI Startup Fund, NVIDIA, and notable individuals like Jeff Bezos, underscores the industry’s confidence in Figure’s mission to address labor shortages in physically demanding sectors through humanoid robotics. The company, which has rapidly achieved unicorn status, plans to accelerate the commercial deployment of its robots, aiming to enhance efficiency in warehouse and retail environments.
In a strategic move to bolster its technological capabilities, Figure has also entered into a collaboration agreement with OpenAI, leveraging the AI leader’s advanced models to power its robots’ cognitive functions. This partnership highlights a significant step towards integrating sophisticated AI with robotic technology, potentially transforming everyday tasks and operations in various industries. With endorsements from high-profile investors and a promising collaboration with OpenAI, Figure is positioning itself at the forefront of the robotics and AI revolution, signaling a future where humanoid robots could become a common aspect of the workforce.
-
Ideogram Secures $80M Series A Led by a16z to Pioneer Next-Gen AI Image Generation
Ideogram, an emerging competitor in the AI image generation space, has secured an $80 million Series A funding round led by Andreessen Horowitz (a16z), marking a significant vote of confidence from one of the tech industry’s leading venture capital firms. This investment propels Ideogram’s valuation and resources, enabling further development of its innovative image-generating model, Ideogram 1.0. The model boasts advanced features such as state-of-the-art text rendering, unparalleled photorealism, prompt adherence, and a unique “Magic Prompt” feature, setting a new benchmark in the AI-driven creative landscape. Available for users to explore freely, Ideogram aims to democratize high-quality image creation with its user-friendly platform and diverse subscription tiers, catering to a wide range of creative needs.
Beyond its technological advancements, Ideogram’s strategic direction is further strengthened by the addition of Martin Casado, a General Partner at a16z, to its board. The company’s vision has attracted a broad spectrum of investors, including Index Ventures, Redpoint Ventures, and Pear VC, signaling robust industry support for Ideogram’s mission to redefine AI-generated imagery. Despite the evolving competitive landscape, with major players like Midjourney and OpenAI’s DALL-E 3 enhancing their text and typography capabilities, Ideogram continues to distinguish itself through its commitment to innovation, user preference, and a suite of features designed to enhance the creative process. This blend of technological excellence and strategic partnerships positions Ideogram as a formidable force in the AI art and image generation domain.
-
SambaNova Unveils Samba-1: A Trillion-Parameter AI Model Revolutionizing Enterprise AI
SambaNova Systems has introduced a groundbreaking large language model (LLM) named Samba-1, boasting an unprecedented one trillion parameters. Unlike conventional single models, Samba-1 utilizes a Composition of Experts architecture, integrating over 50 high-quality AI models. This innovative approach allows for extensive customization and fine-tuning to meet specific enterprise needs, offering a versatile solution for a wide range of applications. SambaNova, primarily known for its advancements in AI hardware, including the SN40L AI chip, aims to redefine the landscape of LLMs with Samba-1. This model is a part of the SambaNova Suite, designed to facilitate the easy customization and deployment of models for organizations, streamlining the process of achieving high performance and scalability in AI applications.
The Composition of Experts architecture of Samba-1 enables a dynamic and efficient interaction among its constituent models, allowing for a seamless integration of responses and inputs across different models. This method stands in contrast to the predetermined model chaining seen in technologies like LangChain, offering a more flexible and adaptive approach to handling complex prompts and generating diverse perspectives. Furthermore, SambaNova emphasizes the security and privacy advantages of this architecture, ensuring that each model remains securely trained on its dedicated dataset. Despite its vast parameter count, Samba-1’s design ensures that only the necessary components are activated in response to specific prompts, leading to significant improvements in efficiency, power consumption, and bandwidth usage. This approach not only enhances the performance of AI applications but also allows enterprises to develop proprietary models tailored to their unique data and requirements, marking a significant step forward in the customization and deployment of enterprise-grade AI solutions.
-
Inkitt Raises $37M to Forge the Future of AI-Powered Publishing and Personalized Storytelling
Inkitt, a startup leveraging AI to unearth potential bestsellers from self-published stories, has successfully raised $37 million in a Series C funding round, aiming to become the 21st-century equivalent of Disney. The platform encourages aspiring authors to publish their stories, which are then analyzed using AI and data science to identify the most promising narratives for further development and distribution through its app, Galatea. With a user base of 33 million and numerous bestsellers under its belt, Inkitt plans to use the new funds to diversify its content offerings. This includes utilizing AI to generate stories from original ideas, creating personalized fiction for readers, expanding into games and audiobooks, and producing more AI-generated video content.
Inkitt’s vision extends beyond competing with platforms like Wattpad; it aims to build a multimedia empire centered around its content library. The Series C round, led by Vinod Khosla of Khosla Ventures and supported by previous investors such as NEA, Kleiner Perkins, and Redalpine, brings Inkitt’s total funding to $117 million. The company, now valued at around $400 million, focuses on innovating the delivery of books to match modern consumption habits, including shorter chapters and immersive reading experiences. By conducting A/B tests on various aspects of its stories, Inkitt gathers valuable data to refine its publishing strategy, aiming for a success rate 20 times higher than traditional publishers. As it ventures into AI-generated and personalized stories, Inkitt is experimenting with different LLMs to maintain its edge in creating compelling content, all while navigating the complexities of intellectual property and data privacy.
-
Photoroom Secures $43M Funding, Valued at $500M for AI-Driven Photo Editing Expansion
Photoroom, a Paris-based AI photo-editing application, has successfully secured $43 million in its latest funding round, achieving a valuation of $500 million. The company, co-founded by CEO Matthieu Rouif and CTO Eliot Andres, has become a significant player in the online business sector, attracting both professional and casual users. Despite initial expectations of a higher raise, the finalized amount underscores the company’s strong market position and growth potential. Photoroom distinguishes itself in a crowded field, competing against entities like Picsart and Pixelcut, by processing approximately 5 billion images annually and surpassing 150 million app downloads. This funding round, led by Balderton Capital with contributions from new investor Aglaé and returning investor Y Combinator, raises Photoroom’s total funding to $64 million.
The investment will fuel Photoroom’s expansion, including doubling its workforce from 50 to 100 employees by year’s end, amidst a broader tech industry trend of layoffs. Unique in its approach, Photoroom invests heavily in R&D and infrastructure, developing proprietary AI models from scratch. This requires significant compute power and image rights acquisitions, driving the need for more technical talent to enhance model efficiency and functionality. Photoroom’s commitment to innovation is evident in its latest feature, Photoroom Instant Diffusion, designed to standardize product photography for a professional studio look, alongside other AI-driven tools for background generation, scene expansion, and bulk image processing. Balderton Capital praises Photoroom’s user-centric vision and generative AI capabilities, highlighting the company’s leadership in the evolving AI landscape.
-
StarCoder 2: Revolutionizing AI-Powered Code Generation Across GPUs
StarCoder 2, the latest innovation in AI-powered code generation, has been released, offering a suite of models designed to run on a wide range of consumer GPUs. Developed through a collaboration between Hugging Face and ServiceNow, with Nvidia joining as the newest supporter, StarCoder 2 aims to provide a more accessible and less restrictive alternative to existing code generators like GitHub Copilot and Amazon CodeWhisperer. The family of models includes a 3-billion-parameter model trained by ServiceNow, a 7-billion-parameter model by Hugging Face, and a 15-billion-parameter model by Nvidia, all designed to enhance coding efficiency without compromising on speed or quality.
Trained on a significantly larger and more diverse dataset than its predecessor, StarCoder 2 boasts improved performance and lower operational costs. It can be fine-tuned in just a few hours using GPUs like the Nvidia A100, making it suitable for developing applications such as chatbots and personal coding assistants. Despite concerns over security vulnerabilities and code sprawl associated with AI code generators, StarCoder 2 offers a promising solution with its BigCode Open RAIL-M 1.0 license, designed to promote responsible use while allowing for a wide range of applications. As the AI landscape continues to evolve, StarCoder 2 represents a step towards more ethical and transparent code generation, with its open-source nature and commitment to training transparency.
-
Adobe Unveils Project Music GenAI Control: A New Frontier in AI-Powered Music Creation
Adobe has introduced Project Music GenAI Control, a groundbreaking platform at the Hot Pod Summit in Brooklyn, designed to revolutionize music editing and creation. This innovative tool allows users to generate audio tracks from text descriptions or reference melodies, offering extensive customization options within the same workflow. Users can modify aspects such as tempo, intensity, and structure, or even extend a track to any desired length, enabling the remixing of music or the creation of endless loops. Currently in the research phase and developed in collaboration with the University of California and Carnegie Mellon, Project Music GenAI Control is yet to be released to the public and lacks a user interface at this stage.
The emergence of GenAI music tools, including Adobe’s latest offering, raises significant ethical and legal questions, especially as AI-generated music becomes increasingly indistinguishable from authentic compositions. Issues of copyright infringement and the legal status of “deepfake” music created by GenAI tools trained on copyrighted content are at the forefront of industry discussions. Adobe is addressing these concerns by developing its GenAI tools with licensed or public domain data and is exploring watermarking technology to identify audio produced by Project Music GenAI Control. Despite these challenges, Adobe remains optimistic about the coexistence of AI-generated music and traditional compositions, anticipating the emergence of new musical ideas and collaborations.
-
Morph Studio Revolutionizes Filmmaking with Stability AI-Generated Video Clips
Morph Studio has launched an innovative AI filmmaking platform, enabling users to craft movies by stitching together video clips generated by Stability AI. This platform, which also incorporates Morph’s own text-to-video model, offers a storyboard-like interface where users can input text prompts to create and edit scenes, forming a unified story. This collaboration with Stability AI marks the beginning of Morph’s plan to offer a selection of generative video models, including its proprietary model, to enhance user creativity. The platform’s workflow allows for generating, editing, and cross-cutting scenes, with the option for users to share their production workflows within Morph’s creator community. This community aspect enables others to replicate and adapt these templates by altering the AI prompts.
Founded in 2023 by Xu Huaizhe and a team of computer vision PhD dropouts from the Hong Kong University of Science and Technology, Morph Studio aims to redefine the filmmaking process. By integrating filming, editing, and post-production into a single, AI-driven workflow, Morph Studio simplifies film production, allowing for immediate shot regeneration and adjustments. With a focus on building a vibrant user community and customizing its AI model to meet creators’ needs, Morph Studio seeks to establish a unique position in the competitive landscape of AI-powered video editing tools. Currently, Morph Studio has a team of about 10 employees and has raised $2.5 million in funding from Baidu Ventures, signaling its commitment to evolving the film production platform and fostering a new era of AI-generated content creation.
-
Apple to Forge Ahead in GenAI, Promises Breakthroughs in 2024, Says Tim Cook
Apple CEO Tim Cook has announced that the company is set to “break new ground” in the field of Generative AI (GenAI) within the year. This declaration was made during Apple’s annual shareholders meeting, amidst news of the company discontinuing its ambitious electric vehicle (EV) project, with some staff from the EV initiative being redirected to various GenAI projects. Despite Apple’s historically cautious approach to GenAI, compared to its Big Tech counterparts, Cook’s statement signals a more assertive move into the domain. Apple has been integrating GenAI internally and is known for its deliberate pace in deploying customer-facing technologies. Recent developments suggest Apple’s GenAI efforts could soon enhance consumer products and services, including upgrades to Siri and the Spotlight search tool to handle more complex queries and interactions.
Apple’s exploration into GenAI extends to potential features for automatically generating presentation slides in Keynote, playlists in Apple Music, and coding suggestions in Xcode. These advancements may be introduced in upcoming versions of iOS, macOS, and iPadOS, expected to be showcased at Apple’s Worldwide Developer Conference this summer. The company’s increasing engagement with GenAI is also evident in its contribution to academic and technical papers on the subject, alongside the release of open-source models and tools for GenAI-powered software development. With a reported annual investment of $1 billion to advance its GenAI capabilities, including the development of a proprietary large language model and an internal chatbot, Apple is poised to make significant strides in GenAI, potentially culminating in enhanced features for the forthcoming iPhone 16 models.
About The Author

Bogdan Iancu
Bogdan Iancu is a seasoned entrepreneur and strategic leader with over 25 years of experience in diverse industrial and commercial fields. His passion for AI, Machine Learning, and Generative AI is underpinned by a deep understanding of advanced calculus, enabling him to leverage these technologies to drive innovation and growth. As a Non-Executive Director, Bogdan brings a wealth of experience and a unique perspective to the boardroom, contributing to robust strategic decisions. With a proven track record of assisting clients worldwide, Bogdan is committed to harnessing the power of AI to transform businesses and create sustainable growth in the digital age.
Leave A Comment