iFLYTEKToday, a major update was released, iFlytek SparkLarge ModelV3.5upgrade, not only launchedThe firstLong text, long image, long voice model,firstThe multi-emotional super-anthropomorphic synthesis technology was introduced to the market, and the Spark Intelligent Agent Platform was launched simultaneously. This series of innovative measures is aimed at providing more powerful technical support for bidding and contract applications.
Liu Qingfeng, chairman of iFLYTEK, proudly stated at the press conference that the Spark Big Model’s ability in general long text processing has reached that of GPT-4Turbo in April.up to dateThe long text version is at the level of 97%, which is an impressive achievement. What is more worth mentioning is that in the vertical field knowledge question answering task, the overall performance of the Spark Big Model even surpassed GPT-4Turbo, demonstrating its excellent performance and strong application potential.
In addition, the Spark app has also been enthusiastically sought after by a large number of users.up to dateStatistics show that as of today, the number of downloads of the Spark app on the Android platform has reached 96 million. This figure not only proves the popularity of iFlytek products, but also reflects the strong market demand for intelligent voice technology.
It is reported that iFlytek Spark Big Model is a new generation of cognitive intelligence big model launched by iFlytek, which has cross-domain knowledge and language understanding capabilities. The big model can understand and execute tasks based on natural dialogue, and provides multiple capabilities, including language understanding, knowledge question and answer, logical reasoning, math problem solving, and code understanding and writing. It has the following seven capabilities:
1. Multi-modal understanding: Upload image material, and the large model completes recognition and understanding, returning an accurate description of the image.
2. Visual question answering: Upload pictures and respond to users’ questions, and the big model will answer them.
3. Multimodal generation: Generate synthetic audio and video that meet the user's expectations based on the user's description.
4. Virtual human video: Describe the desired video content, integrate AI virtual humans, and quickly generate matching videos.
5. Large model speech recognition: supports mainstream languages and leads the world, improves speech recognition accuracy, supports 37 languages, and realizes automatic language judgment and specified language recognition.
6. Large model speech synthesis: Provides super-humanized speech synthesis capabilities and achieves high-accuracy speech synthesis.
7. Large model code: Achieve code understanding and generation capabilities, reaching the level of 96%.
iFlytek Spark Big Model can also be accessed through API, quickly acquiring cross-domain knowledge and powerful natural language understanding capabilities. At the same time, Spark Assistant provides a variety of intelligent assistant applications, such as PPT outline assistant, business copywriting generation, mock interview assistant, etc., so that every scenario can find a big model application that can be used out of the box. In addition, the plug-in market and native applications also provide developers and users with more functions and tool options to jointly build the iFlytek Spark Big Model ecosystem.
Experience address:https://www.1ai.net/5211.html