Xunfei Xinghuo API
Up to 10 million tokens! iFlytek Spark large model API
iFLYTEK StarFire API Official Website, StarFire Large Model API Pricing, Application, Usage, Documentation
Introduction to the iFLYTEK StarFire Cognitive Large Model
The iFLYTEK StarFire Cognitive Large Model possesses cross-domain knowledge and language comprehension capabilities, completing tasks such as Q&A dialogues and literary creation. It continuously learns and evolves from massive text data and large-scale grammatical knowledge, achieving a full-process closed loop from posing questions, planning questions to solving problems.
iFLYTEK StarFire API Official Website:https://xinghuo.xfyun.cn/sparkapi
>>>Claim a Free Certification Gift Pack, Up to 10 Million Tokens!
1. Interface Description
Note: This interface is available for official use. If you need to apply for use,>>>please click here to go to the product page .
StarFire API Pricing
Pricing Information:
Service Engine Unit Price
iFLYTEK StarFire Cognitive Large Model V1.5 0.18 Yuan/10,000 Tokens
iFLYTEK StarFire Cognitive Large Model V2.0 0.36 Yuan/10,000 Tokens
iFLYTEK StarFire Cognitive Large Model V3.0 0.36 Yuan/10,000 Tokens
Tips:
- Pricing includes both input and output content of the interface
- 1 token is approximately equal to 1.5 Chinese characters or 0.8 English words
- StarFire V1.5 supports [Search] built-in plugins; StarFire V2.0 and V3.0 support [Search], [Weather], [Date], [Poetry], [Words], [Stock] six built-in plugins
- StarFire V3.0 now supports Function Call functionality.
#
Application Scenarios
Language Comprehension, Content Creation, Knowledge Q&A, Logical Reasoning
Mathematical Ability, Code Understanding and Writing, Multi-modal Interaction Look Forward to More
iFLYTEK StarFire Cognitive Large Model Web API Documentation
For details, please refer to here:
https://www.xfyun.cn/doc/spark/Web.html
Product Advantages
Fast response, efficient processing
Adopting a streaming interface design, the first frame response can reach the millisecond level. With efficient algorithms and architecture, it can process a large number of user requests in a very short time
Diverse Scenarios, Continuous Evolution
Provides diverse capabilities including language understanding, knowledge Q&A, code writing, logical reasoning, mathematical problem-solving, etc., and continuously learns and evolves from massive data and knowledge
Flexible Application, Personalized Customization
Provides rich parameter settings to achieve personalized model experiences. For enterprise-level customized needs, dedicated model solutions can be provided
Stable Service, Secure and Reliable
Cloud services reach the level of national security standard, adopting a public cloud architecture with a billion-level traffic, combined with multiple means such as data encryption and access control to ensure user privacy security
QA Q&A
Empower customers, improve business efficiency and competitiveness, create greater commercial value
How to use the multimodal generation capabilities of the large model?
Currently, StarFire V1.5 and V2.0 only support text-to-text interaction. The multimodal capabilities will be provided as a separate API in the future, please stay tuned!
If you need to experience multimodal interaction now, you can go to SparkDesk for a try.
How to combine historical conversation information with the large model interface?
When calling the API, if you want to implement Q&A combining historical conversation information, you need to developers to concatenate the latest questions and historical conversations in a fixed format and send them to the interface when calling the interface. For details, please click to jump to the documentation
Why do I get an "Unauthorized" error when calling the interface?
1. Please check if your APPID has authorization; 2. Please check in the console if the APPID and APIKey and APISecret match; 3. Please check if the APIKey and APISecret are filled in the wrong way in your code. If all the above are correct, you can submit a ticket to feedback your problem.
What are the limitations on interface input?
The input limitations will gradually increase with customer needs in the future, and you can check the content field in the interface request part of the development documentation payload.message.text.