iFlytek MaaS

What is Xunfei Xingchen MaaS?

iFlytek MaaS is a one-stop large-scale model fine-tuning and service platform launched by iFlytek, integrating model inference APIs, zero-code fine-tuning, and developer subscriptions. The platform aggregates mainstream large models such as DeepSeek, Qwen, GLM, Kimi, MiniMax, and iFlytek Spark, providing dual-mode model customization: one-line fine-tuning and professional fine-tuning. For developers, it offers the Astron Coding Plan subscription service, replacing pay-as-you-go billing with a fixed monthly fee, covering the entire chain of needs from model access and fine-tuning to AI programming.

The main functions of iFlytek StarMaaS

• Model Inference : Provides real-time inference services and batch inference capabilities, supporting high-concurrency API calls and one-click deployment.
• Model fine-tuning : Supports zero-code fine-tuning, Notebook development, and model distillation to meet the needs of different developers.
• Model Hosting : Provides proprietary model hosting and pre-built model hosting services to achieve efficient model deployment and management.
• Model evaluation : Supports multi-dimensional performance evaluation and provides multiple scoring modes for factual and open-ended questions.
• Dataset Management : Provides preset and custom datasets, and supports data upload and version management.
• Data cleaning and augmentation : Use data augmentation tools to generate more training samples, improving data quality and model performance.
• Question and answer pair extraction : Supports automatic segmentation and extraction of Q&A pairs that conform to the platform’s specifications from text files or website links.
• Prompt Project : Provides preset Prompt templates and customization features to help developers optimize instruction design.
• Experience Center : Supports online comparison of models before and after fine-tuning, allowing for quick verification of model performance.
• Large Model Application Platform : Provides a one-stop development environment for large model applications, lowering the barrier to AI application development.
• Model Square API Service : Provides inference APIs for popular models such as DeepSeek-V4-Pro , Spark-X2-Flash , Qwen3.6-35B-A3B , GLM-5.1 , and Spark X2 , supporting text generation and multimodal understanding. Some models offer free inference quotas.
• Astron Coding Plan : A monthly subscription coding service for developers, integrating multiple model foundations and accessing mainstream programming tools such as Cursor , Claude Code , and OpenClaw through the OpenAI/Anthropic dual protocol.
• In short : Designed for users with no prior experience, it provides preset scenarios such as roles, categories, and creation. Simply input a scenario description to quickly generate your own model, without the need for training data.
• Professional model fine-tuning : Supports uploading custom datasets (json/jsonl/csv format, ≥2000 records recommended), and allows for in-depth fine-tuning through a complete toolchain, with real-time monitoring of loss function and accuracy.

iFlytek Star MaaS Core Advantages

• Model ecosystem aggregation : One platform integrates mainstream domestic and international models, and users can switch bases as needed without having to connect to multiple vendor APIs separately.
• Dual-mode fine-tuning threshold : from zero-threshold quick verification with a single sentence to enterprise-level in-depth customization with professional fine-tuning, covering user needs at all stages.
• Data Engineering Closed Loop : Unique question-and-answer pair extraction and data augmentation tools automatically transform unstructured documents into training data, significantly reducing data preparation costs.
• Developer subscription innovation : Coding Plan replaces pay-as-you-go billing with a fixed monthly fee, completely avoiding the risk of overdue payments, and is compatible with the mainstream AI programming tool ecosystem.

How to use iFlytek MaaS

• Registration and Access : Visit the iFlytek MaaS official website https://training.xfyun.cn/modelSquare , complete account registration and login, obtain the API key, and quickly access the platform through the OpenAI compatible interface.
• Choose a model : Browse 70+ preset models in the model marketplace and select the appropriate text, image, or multimodal model based on your scenario.
• Data preparation : Upload a custom dataset or use a pre-built dataset, and optimize data quality through data cleaning, augmentation, and question answering.
• Model fine-tuning : Choose zero-code fine-tuning or Notebook development, configure the parameters, submit the training task, and view the training progress and metrics in real time.
• Deployment and Invocation : After training is complete, publish the model service, integrate it through API calls or batch inference, and verify the effect in the experience center.
• Application building : Connect to open source applications such as OpenClaw and OpenCode, or develop a dedicated AI Agent for large model application platforms.