The latest version 2.6 of WallFacer’s MiniCPM-V series has rapidly climbed to the Top 3 on GitHub and HuggingFace trends, surpassing ten thousand stars. Since its release in February, it has accumulated over a million downloads, becoming a benchmark for on-device model capabilities. MiniCPM-V2.6 achieves performance enhancements for on-device multimodal models with 8 billion parameters, including real-time video understanding, multi-image joint understanding, and multi-image in-context learning, with a quantized backend memory of only 6GB and an inference speed of up to 18 tokens.

Google has released its latest AI masterpiece, Gemini 1.5 Pro, offering an experimental version 0801 through Google AI Studio and the Gemini API. This model leads the LMSYS leaderboard with an ELO score of 1300, surpassing OpenAI's GPT-4o and Anthropic's Claude-3.5 Sonnet. Gemini 1.5 Pro excels in multilingual tasks, mathematics, coding, and visual tasks, featuring a context window of 2 million tokens.

Shusheng · Puyu Lingbi (InternLM-XComposer) Version 2.5 was developed by the Shanghai Artificial Intelligence Laboratory, focusing on long context input and output capabilities, operating smoothly within a length of 96K, and trained with 24K interleaved image-text data. Key upgrades include: high-resolution image understanding, fine-grained video understanding, and multi-turn multi-image dialogue. In application, it can create web pages and write high-quality text-image articles. Evaluations show it surpasses state-of-the-art open-source models across 16 benchmark tests and performs at par with key tasks compared to GPT-4V and Gem.

Meta is about to make a big move! They are set to launch an open-source language model called Llama3405B, which is not only their largest model to date but also the largest open-source language model in history. This behemoth, with an astonishing 405 billion parameters, can effortlessly navigate between images and text, completely revolutionizing the old ways that could only handle text.Key Highlights: Meta will release Llama3405B on July 23rd, a multimodal model with 405 billion parameters. Dec

In the world of AI, making machines understand videos is much harder than understanding images. Videos are dynamic, with sound, movement, and a myriad of complex scenes. In the past, AI viewed videos as if they were reading ancient scrolls, often leaving them baffled. But the introduction of VideoPrism might change everything. It's a video encoder developed by Google's research team that can achieve state-of-the-art levels on a variety of video understanding tasks with a single model. Whether it

Quick Hands released a big move today by opening its in-house image generation model——“Kolors”. This is not an ordinary model; it has been trained on tens of billions of text-image pairs, equipped with a General Language Model (GLM) as a text encoder, supporting bilingual Chinese and English prompts, and can handle contexts up to 256 tokens. Key Features of Kolors:Bilingual Support:Utilizes the General Language Model (GLM) as a text encoder, enabling the model to not only master English but

Keling AI has made another big splash recently, not only launching a web version but also releasing several major new features. The launch of the Keling web version, a comprehensive improvement in image quality, the addition of start and end frame and camera control functions, as well as the extended open length for text-to-video up to 10 seconds. Key Updates:Basic Model Upgrade: Keling AI's basic model has undergone significant upgrades, now capable of generating videos with higher resolutions,

In the world of AI, there are always some innovative things that catch one's eye, such as the recently popular domestic animation video AI—YoYo. It allows countless anime enthusiasts to unleash their creativity, making characters that were once only in imagination instantly accessible. Imagine, with just a few taps, you can input a few keywords or upload an image, and YoYo can generate a lifelike anime video for you. This is no longer just a simple animation of images; it is a feast of both vis

In the dazzling galaxy of AI, Leap Star Company has emerged as a shining star with its innovative multimodal models. At the WAIC conference, they showcased three uniquely crafted AI models, highlighting their multimodal capabilities. Step-2: The MoE model with 10 billion parameters is currently available for experience upon application. Step-1.5V, a billion parameter multimodal model, not only demonstrates extraordinary talents in image understanding but also opens up new horizons in video under

Google is serious this time! According to the latest leak, the Google Pixel 9 series is set to bring a series of eye-catching AI features. Highlight One: Add Me Feature Imagine your group photos always have regrets due to someone blinking or not smiling. Now, with the Add Me feature, these regrets will be a thing of the past. It can capture each person's best expression in a group photo, even merging expressions from different photos into a single group shot, ensuring everyone appears at their b

The French independent non-profit AI research lab Kyutai has launched a voice assistant called Moshi, which is a revolutionary real-time local multimodal foundational model. This innovative model imitates and surpasses some of the functionalities demonstrated by OpenAI's GPT-4o released in May in certain aspects.Product Entry: https://top.aibase.com/tool/moshi-chat Moshi is designed to understand and express emotions, capable of conversing in different accents, including French. It can simultane

Developed jointly by researchers from the University of Hong Kong and Bytedance, LlamaGen is an image generation method based on the autoregressive model Llama, which has shown the potential to surpass traditional diffusion models in the field of image generation. The open-source release of LlamaGen quickly garnered nearly 900 stars on GitHub, acknowledging this achievement. This result not only proves the competitiveness of autoregressive models in image generation but also brings new vitality

Lately, a user named Home Assistant on the Internet has come up with a new twist on GPT-4V. He boldly connected GPT-4Vision directly to his home camera, allowing the AI to observe him around the clock without any blind spots, and the resulting video has been viewed over a million times on X! Imagine walking around your home and suddenly realizing you're being watched 24/7 by an AI that covers every angle. It might feel a bit like a "lab rat," but this foreign netizen takes it in stride. He think

TRANSAGENTS, a virtual translation publishing company developed by Tencent's AI Lab specifically for literary translation, has officially launched and is now available for free online experience. TRANSAGENTS is a novel multi-agent framework designed for translating long literary content. It simulates a real translation company through virtual roles, including senior editors, junior editors, and translators, each with unique responsibilities and tasks. This multi-role collaboration model makes th

Lately, young people in the United States have experienced a collective "heartbreak." The AI chatbot Character AI, which they had become obsessed with, suddenly seemed boring. The reason behind this might be due to some instructions from Google or Meta. Character AI was once the most popular social app among Generation Z in the United States. Its chatbot model handled up to 20,000 queries per second at its peak, which is one-fifth of Google's search volume. However, now, this AI that once kept t

Recently, Alibaba's Entertainment and Culture Division has launched its first makeup and costume design large-scale model called Shenni Ningchang. This model specializes in the creative design of traditional costumes for TV dramas, featuring three major functions: styling design, makeup and hair design, and pattern design. By setting conditions such as dynasty, gender, and identity, and inputting relevant keywords, a makeup and costume image can be generated in just 2 minutes. This technology ha

Recently, Sun Yat-sen University and the BytePlus Digital Human Team made a big splash by proposing a virtual try-on framework called MMTryon. This isn't just any technology; with just a few images of clothes and a few instructions on how to wear them, you can generate a model wearing the clothes with a single click, and the quality is exceptionally high.Imagine selecting a coat, a pair of pants, and a bag, and then "wham!" they automatically appear on the human figure. Whether you're a real per

In the field of artificial intelligence, code generation and review have always been critical battlegrounds for technological advancement. OpenAI has recently launched CriticGPT, a model based on GPT-4, specifically designed to review code generated by ChatGPT and identify errors within it. The introduction of this innovative tool marks a significant step forward in artificial intelligence's capabilities in self-monitoring and error detection.

As the flames of AI battle move towards the education sector, iFLYTEK unveils its Spark Large Model V4.0, marking the start of a contest between intelligence and wisdom. This is not only a leap in technology but also a disruption of traditional educational models. At this launch event themed 'Your Understanding AI Assistant', iFLYTEK issued a challenge to GPT-4 Turbo, vowing to compete in the battlefield of AI.

The 'invasion' of AI into university exam halls has quietly occurred. Research from the University of Reading in the UK has revealed a staggering phenomenon: in actual university exams, 94% of AI-generated answers went unnoticed by teachers and were not detected. Even more shocking is that these AI 'students' achieved scores that were significantly higher than human students 83.4% of the time. This seems to signal that AI not only shows potential in replacing human jobs but is also beginning to surpass university students in cognitive tasks.