OpenAI CEO Sam Altman announced that there will be a 12-day OpenAI event coming up.

Every working day, there will be a live broadcast, including the latest technology or product demos. Expected releases will include the highly anticipated text-to-video AI tool Sora from OpenAI and a new inference model.

Expected potential releases:

Release of the text-to-video model Sora: Expected to showcase more Sora video creations and officially release the text-to-video model Sora.

Orion model demonstration: According to previous reports, OpenAI plans to introduce a new model codenamed "Orion", which is expected to be 100 times more powerful than the GPT-4. The livestream is expected to demonstrate the model.

Preview of the upcoming 2025 Agent product: OpenAI plans to launch an AI assistant named "Operator" that will be able to handle complex tasks such as code generation, travel booking, and even automated e-commerce shopping. Expected to be officially released in 2025, but this livestream may bring the first appearance.

AI application growth is accelerating month-on-month, with continuous overseas catalysis.

As of October data, according to the AI product ranking, AI applications represented by ChatGPT have gradually recovered since Q3 and maintained a high level of 17.16% in October. The user base is steadily growing, and user usage habits are gradually strengthening.

In the past month, overseas application layer representative companies have successively exceeded expectations in performance, and AI advancement is steadily moving forward as well.$Applovin (APP.US)$、$Salesforce (CRM.US)$、$Palantir (PLTR.US)$、$Duolingo (DUOL.US)$、$Innodata (INOD.US)$Represented by等。

Industrial end steadily advancing, with a constant stream of new Agent products.

1) Although the application end has not yet achieved an explosive breakthrough, the industrial end has been steadily advancing.$Microsoft (MSFT.US)$ Copilot Studio's autonomous intelligent body function has now launched the preview version AXON 2.0 to provide power for AppDiscovery, salesforce has officially announced the full launch of Agentforce, Duolingo has released the function of Lily video calls for AI characters, and the product is steadily iterating.

With the increasing ability of large models to solve complex tasks, the deployment of Agents is increasing. Anthropic, microsoft, salesforce, and OpenAI have all made key innovations and are expected to gradually be implemented.

In the medium term, domestic application vendors are expected to achieve resonance between AI promotion and the fundamental turning point.

Since the beginning of this year, there has been some volatility in the external environment, and downstream IT spending may change, putting pressure on the performance of computer software application vendors.

Since the end of September, several stimulus policies have been implemented, and it is expected to turn better by 2025. The logic of repairing computer investment is expected to be realized, and the inflection point of software vendors' performance may gradually appear.

Meanwhile, in the future, with the significant improvement in the accuracy, logic, and memory of large models, AI applications will greatly enhance work efficiency, and customer stickiness and willingness to pay will continue to increase.

Recently, the overseas catalysis of AI+ applications continues, and AI continues to empower the financial results of many technology companies beyond expectations, with continued release of heavyweight AI+ products. AI+ application companies represented by salesforce, Applovin, palantir, and others have recently achieved significant gains.$Shopify (SHOP.US)$Continuously bullish on the progress of AI model iterations and accelerating the development of AI applications in advertising, e-commerce, education, film, and television directions.

Asia Vets APP learned that the narrative logic of AI applications has been further validated, and commercial progress is gradually being implemented.

GTJA released a research report stating that as the AI application process accelerates and scenarios continue to emerge, coupled with the dual drive of cloud applications and terminal intelligence to drive industrial upgrades, AI track companies are ushering in important development opportunities. It is recommended to focus on the following directions: Leading application companies driven by product strength, especially those companies that have formed technological barriers and user stickiness in existing fields; Enterprises with first-mover advantages in cloud application innovation and terminal intelligence upgrades, as well as companies that have achieved commercialization in high-barrier scenarios such as medical, financial, and educational fields; Companies with core technology and resource accumulation in the field of Chinese language corpus construction. These companies will gain sustained competitive advantages in the process of AI application ecosystem improvement.

AI related Hong Kong stocks:

$MOBVISTA (01860.HK)$ : The company plans to release its third-quarter results on November 15. Its advertising platform, Mobvista, performed well in the latest report, especially achieving new breakthroughs in the dual-end game list and the Android end. Open-source Securities is bullish on its platform's scale growth and multi-vertical expansion as a mid-term strategic goal.

$YIDU TECH (02158.HK)$ : MediTech's self-developed large model is a professional large language model for multiple scenarios in the medical vertical field and has successfully adapted to the Ascend AI in the medical field. Currently, its performance in many medical specific task scenarios such as differentiated diagnosis, basic medicine, general practice medicine, clinical internal medicine, clinical surgery, and professional qualification examinations exceeds GPT3.5. In data structuring tasks, the accuracy and recall rate are also much higher than similar parameter large models. The company positions the large model as the new generation "medical super-brain," and its applications have already landed on the consumer (C) and business (B) ends. For consumers, MediTech's large model can provide professional medical-level personal services and has been piloted in the "Huimin Insurance" field; The Jiangsu provincial-level "Jiangsu Medical Benefit Insurance No. 1" is the first to access the MediTech large model, providing 24/7 all-day health management services.

$FOURTH PARADIGM (06682.HK)$In the first three quarters of 2024, the fourth paradigm's three major business lines performed differently. Among them, the core business, the highest revenue business - the Prophet AI platform, maintained a high growth rate during the reporting period, achieving actual revenue of 2.171 billion yuan, a year-on-year increase of 50.2%. The proportion of total revenue further increased to 67.5%.

$AINNOVATION (02121.HK)$The company's continuous driving force for "AI + manufacturing" is sufficient. Since 2023, the company has implemented the dual-tower development of AI 1.0 and AI 2.0. AI 2.0 gradually launched the application and pre-research of the ChatX series of vertical large models, and at the end of March this year, released the AnnoGC-75B industrial large model (75 billion parameters), adding multimodality on the basis of the 15B large model. Currently, ChatX includes 5 major landing products: ChatBI, ChatDoc, Chatvision, ChatCAD, and Chatrobot. The company's 9 major segments cover industrial software, digital software, industrial logistics, intelligent equipment, and industrial sustainability in five major scenes. ChatBI has landed in the industrial software and digital software scenes of the food and beverage intelligent manufacturing field. ChatDoC and Chatvision have landed in the industrial sustainability scene. ChatCAD has greater imagination space in the CAD industrial design field. Through large models, it can rewrite traditional CAD design patterns and directly carry out CAD design through text and language, significantly improving CAD design efficiency. It is expected that the company will start the landing practical work of ChatCAD this year. ChatRobot is driving industrial robot automatic task scheduling with large models, driving and implementing work deployment. The company has done on-site demonstrations at the large model release conference. Although still in the pre-research verification stage, it has preliminary industrial application prospects and great potential.

$FENBI (02469.HK)$In 2024, fenbi is driving the application of large AI models and took the lead in releasing AI teacher products in the industry to meet users' diverse learning needs and share innovative development results with users for the first time. Fenbi uses the divide-and-conquer method to enhance the model's understanding of problems by breaking down complex issues into simple enough tasks, which is far better than directly handing over complex tasks to the model. According to previously released data, fenbi assistants and teachers have provided approximately 0.5 billion conversation templates, with the fenbi question bank accumulating 3.748 billion practice sessions and users completing 61.3 billion questions. Based on this, fenbi has subdivided user intentions into 11 major categories and 42 subcategories, processing over 0.3 million conversation data to enhance the AI teacher's ability to handle complex dialogues. Fenbi is also developing more diverse AI scenarios, including the Essay AI teacher that will be launched this year, along with AI teachers for interviews, civil service and teacher projects in preparation.

Editor/ping