Welcome to the AI Daily column! This is your daily guide to exploring the world of artificial intelligence, where we present the hottest topics in the AI field every day, focusing on developers to help you understand technological trends and innovative AI product applications.
Fresh AI Products Click to Learn More: https://top.aibase.com/
📰🤖📢AI News
OpenAI Releases GPT-4-Turbo Official Version with Image Reading Capabilities
AiBase Highlights:
⭐ The official version of GPT-4-Turbo is now fully available, accessible via the model "gpt-4-turbo," with the latest version being "gpt-4-turbo-2024-04-09".
⭐ GPT-4-Turbo is a model with visual capabilities, capable of processing 128k context.
⭐ Pricing remains unchanged: $10.00/1 million tokens for input, $30.00/1 million tokens for output, and $0.00085/image for image reading.
API Information: https://platform.openai.com/docs/models/continuous-model-upgrades
Pricing Information: https://openai.com/pricing
Related Restrictions: https://platform.openai.com/docs/guides/rate-limits/usage-tiers?context=tier-five
Significant! GPT-4 API Now Fully Available for Use
AiBase Highlights:
⚙️ GPT-4 boasts enhanced creativity, supporting the generation, editing, and collaborative iteration of creative and technical writing tasks.
🔍 GPT-4 has a longer context processing capability, suitable for creating long-form content, extending dialogues, and document search and analysis.
🏢 Companies like Morgan Stanley have begun applying GPT-4, embedding it into internal bots to assist financial personnel in finding information and saving time.
API Detailed Usage Instructions: https://platform.openai.com/docs/guides/gpt
Google Announces Gemini1.5Pro API with New Features
AiBase Highlights:
🎵 Gemini1.5Pro introduces native audio understanding capabilities without the need to convert to text.
📄 Gemini1.5Pro expands the range of input modalities, supporting video content processing.
💻 Gemini API improvements include system command guidance for model responses and JSON mode.
Details Here: https://developers.googleblog.com/2024/04/gemini-15-pro-in-public-preview-with-new-features.html
Audio Understanding Capabilities: https://github.com/google-gemini/cookbook/blob/main/quickstarts/Audio.ipynb
SD3 to be Released Mid-April! Expected to Include Models of Six Different Sizes
AiBase Highlights:
⭐ SD3 is planned to be officially released mid-April, with two papers on SD3 and SD3Turbo already published.
⭐ SD3 may include up to six different sizes of models to meet various needs.
⭐ SD3 has 8 billion parameters, suitable for RTX4090, generating 1024x1024 images in 34 seconds.
SD3 Paper: https://arxiv.org/abs/2403.12015
SD3Turbo Paper: https://arxiv.org/abs/2403.03206
More SD3 Image Creation Here: https://sd3.art/
Performance Surpasses H100! Intel Releases New Generation AI Chip Gaudi3, Pressuring NVIDIA!
AiBase Highlights:
⭐ Gaudi3 outperforms NVIDIA's H100, making it one of the strongest AI chips.
⭐ Gaudi3's energy efficiency is more than twice that of NVIDIA's chips, with faster speeds.
⭐ Gaudi3 uses advanced 5nm technology, equipped with 128GB HBM2e memory.
Poe Introduces a Message-Based Pricing Revenue Model for AI Robot Creators
AiBase Highlights:
⭐ Poe has launched a message-based revenue model.
⭐ The new revenue model will help incentivize the development of a wider variety of robots.
⭐ Creators can set the price per message for their robots, earning revenue.
Just Released, Mistral AI's Latest Magnet Link! 8x22B MoE Model, 281GB Unlocked
AiBase Highlights:
🌟 Mistral AI has released another magnet link, unlocking the 281GB 8x22B MoE model.
🌟 Mistral AI has stirred up the AI community with a single magnet link.
🌟 The new MoE model supports multiple languages and boasts strong performance, directly competing with GPT-4.
Douyin AI Special Effects Create Miyazaki Hayao's Art Style, Going Viral
AiBase Highlights:
⭐ The film promotion team collaborated with Douyin to launch an AI special effect in the style of Miyazaki Hayao's animations, attracting 2 million users to participate.
⭐ The use of the special effect sparked controversy; supporters believe it pays tribute to Miyazaki Hayao, while opponents think it violates the master's intentions.
⭐ The controversy points out Miyazaki Hayao's resistance to AI animation technology, but the actual video was not created using AI.
Suno Music Dominates Bilibili's Ghost畜 Zone, "Steel Door Laxity" Becomes the Hottest Topic
AiBase Highlights:
🎶 The song "Your Steel Door is Quite Lax" has become a sensation on Bilibili, attracting millions of netizens.
🎤 AI music gains popularity, with Suno software leading the creation of ghost畜 music, achieving breakthrough success.
📈 The AI music comment section is harmonious, with netizens interacting in a humorous and teasing manner, showcasing creative flair.
Apple Releases New Machine Learning Language Model MLLLLM Ferret-UI for Understanding Application UI Interfaces
AiBase Highlights:
⭐️ Ferret-UI is a MLLLM designed specifically for mobile UI screens, with referencing, locating, and reasoning capabilities.
⭐️ It adopts "any resolution" technology to improve the model's understanding accuracy of small objects.
⭐️ Ferret-UI excels in tasks such as detailed description, interactive dialogue, and functional inference through its conversational abilities.
Paper Address: https://arxiv.org/pdf/2404.05719.pdf
Jack Ma Speaks Out on Alibaba's Transformation: AI Era is Just Beginning
AiBase Highlights:
⭐ Jack Ma commends Alibaba's new management's courage, stating the company has returned to a healthy growth trajectory.
⭐ Ma believes Alibaba's core change is the return to a customer value-centric approach.
⭐ Ma predicts that the internet industry will undergo a dramatic transformation in the next three to five years, with the AI era just beginning.
🤖📱💼AI Applications
Baidu Cloud Launches AI Creative Generation Tool "Super Canvas"
AiBase Highlights:
⭐ Unique image large model and intelligent face fusion algorithm.
⭐ Offers a rich selection of backgrounds and clothing options.
⭐ Allows for local redrawing of facial expressions and hair color.
Experience Here: https://top.aibase.com/tool/chaonenghuabu
Effective Free (No Login Required) GPT-3.5 Projects
1⃣ Aurora https://github.com/aurora-develop/aurora
(Various ways to use for free, supports one-click deployment on Render/Vercel, etc.)
2⃣ ChatGPT https://github.com/PawanOsman/ChatGPT
(Supports Docker & Android, very stable)
3⃣ FreeGPT35 https://github.com/missuo/FreeGPT35
(Many parameters, supports Nginx reverse proxy)
Google to Launch New AI Video Tool Google Vids with Multiple Features
AiBase Highlights:
📱 Google Vids integrates multiple features to help users easily create storyboards.
🖼️ AI technology assists in editing content, selecting suitable scenes, images, and music.
🎙️ Users can add voiceovers to videos, enhancing personalization and professionalism.
