Yesterday.AliCloud made a big move, one of their advanced AI model called Qwen-Long suddenly reduced the price, reduced particularly hard, would have used this model to spend the money can now save 97%, 1 dollar can buy the previous hundreds of dollars to buy the service, which is the same as buying afive booksThe New Dictionary is about the same money.
Aliyun,BaiduIntelligent Cloud,ByteDanceSuccessive major announcements willLarge ModelThe cost of using the program has dropped to freezing point, and this "price war"Not only has it reshaped the industry landscape, but it's more broadlyBusinesses and developersbrings unprecedented opportunities.
AliCloud: Qwen-Long "Breaker" with Record Low Costs
AliCloud struck first with API inputs for its GPT-4 level master model Qwen-LongPrice Direct 97%Qwen-Long supports up to 10 million tokens of long text input, which is 400 times more cost-effective than GPT-4. This price cut by AliCloud is undoubtedly a deep water bomb dropped on the market, heralding a revolutionary reduction in the cost of large models.
Baidu Wenshin: Free Admission, Opening the Era of "Zero Threshold" for Big Models
Following in the footsteps of AliCloud's price cuts, Baidu SmartCloud reacted quickly by announcing that its two main models, ERNIE Speed and ERNIE LiteCompletely free of charge, effective immediatelyThis is not only a bold marketing strategy, but also a shock to the whole industry. This is not only a bold market strategy, but also a shock to the whole industry, announcing that large model applications have officially entered the "free era". ERNIE Speed and ERNIE Lite meet the needs of different scenarios with their high-performance and lightweight features, and the free openness will undoubtedly attract a large number of developers to join them and accelerate the emergence of innovative applications. The free opening will undoubtedly attract a large number of developers to join and accelerate the emergence of innovative applications.
Byte Jumping Beanbags: Low Price Pioneers, Accelerates Industry Change
Before Ali and Baidu, byte jumping has been a step ahead, the launch of self-research beanbag big model, pricing is only 0.0008 yuan/thousand tokens, lower than the industry average of 99.3%, a dollar can buy the equivalent of three books of the "Romance of the Three Kingdoms" volume level of text processing capacity. Byte jumping this move is seen as the industry "fig. murderous dictator", significantly lowering the threshold for enterprises to explore AI applications and driving big models from concept to practice.
The Logic Behind the Price War: Accelerating AI Transformation and Driving Value Creation
In the midst of all the price cuts, Liu Weiguang, senior vice president of AliCloud Intelligence Group, pointed out that the purpose of significantly reducing the price of large model inference is to accelerate the explosion of AI applications. He believes that whether it is open source or commercialized models, the public cloud + API model will become mainstream because it can take full advantage of the technological dividend, reduce costs and improve performance, while guaranteeing data security and providing a rich selection of models.
Baidu, on the other hand, emphasized that the value of the big model lies not only in the price, but also in the comprehensive effect, including the speed of response and the wide range of application scenarios. Baidu Wenxin big model of the average daily processing tokens text volume of up to 249 billion, highlighting its strong strength in performance and efficiency.
Industry impact: accelerating innovation, ecological win-win or pressure multiplier?
This round of price wars for large models will undoubtedly greatly stimulate the innovative vitality of B-side enterprises and AI startups, resulting in a significant drop in the cost of AI application development and accelerated landing. However, for small-scale model providers, this may mean unprecedented competitive pressure, forcing them to re-examine their market positioning and business model.
CRE Securities research report pointed out that the cost reduction is to promote the large model into the "value creation stage"The key. Although cost compression may prompt companies to strengthen cost control in the supply chain, the low-cost environment will promote the prosperity of the AI model ecosystem and drive the innovative explosion of end-use applications at the current critical juncture of AI transformation.
By doing this, these big companies are actually telling everyone that they don't have to spend too much money to use AI in the future, and that both big companies and small developers can afford to play with it. In this way, more and more people will use AI to do all kinds of new things, and AI will become more grounded. However, smaller AI modeling companies may have to have a headache because the competition is too fierce. Overall, the world of AI is becoming more lively and hands-on.