OpenSearch LLM-powered AI Chat Edition includes built-in, high-performance large language models (LLMs). You can also add external LLMs from Alibaba Cloud Model Studio to select and integrate the models that best suit your business needs. This improves the diversity and adaptability of the system. This topic describes how to view the list of LLMs and add external LLMs.
Model list
Go to Model Management > LLM to view the list of LLMs. The list includes information such as Model Name, Model Type (Built-in LLM, External LLM, or Custom LLM), and Creation Time. You can also perform operations such as Q&A Test.
Model | Type | Description |
qwen3-235b-a22b | Built-in LLM | A new-generation Qwen large language model. Based on extensive training, Qwen3 makes breakthroughs in inference, instruction following, agent capabilities, and multilingual support. It supports over 100 languages and dialects with powerful multilingual understanding, inference, and generation capabilities. |
qwen-vl-max | Built-in LLM | A Qwen large-scale vision-language model. It supports multimodal retrieval-augmented generation (RAG) for images and videos, providing optimal performance on more complex tasks. |
qwen-vl-plus | Built-in LLM | A Qwen large-scale vision-language model. It supports multimodal RAG for images and videos, delivering the fastest response speed. |
opensearch-qwen | Built-in LLM | A model based on Qwen-14b, fine-tuned with retrieval-augmented training data. |
opensearch-qwen-plus | Built-in LLM | A model based on Qwen-72b, fine-tuned with retrieval-augmented training data. It provides more stable performance for the same questions. |
deepseek-r1 | Built-in LLM | The deepseek-r1 model. It performs better on tasks such as mathematical reasoning. |
deepseek-r1-distill-qwen-7b | Built-in LLM | The deepseek-r1-distill-qwen-7b model from Alibaba Cloud Model Studio. |
deepseek-r1-distill-qwen-14b | Built-in LLM | The deepseek-r1-distill-qwen-14b model from Alibaba Cloud Model Studio. |
opensearch-llama2-13b | Built-in LLM | A model based on llama2-13b, fine-tuned with retrieval-augmented training data. |
opensearch-llama2-13b-turbo | Built-in LLM | An upgraded model based on llama2-13b, fine-tuned with retrieval-augmented training data. |
opensearch-llama2-70b | Built-in LLM | A model based on llama2-70b, fine-tuned with retrieval-augmented training data. |
opensearch-llama3-8b | Built-in LLM | A model based on llama3-8b, fine-tuned with retrieval-augmented training data. |
qwen-turbo | Built-in LLM | The fastest and most cost-effective model in the Qwen series. It is suitable for simple jobs. |
qwen-plus | Built-in LLM | A balanced model in the Qwen series. Its inference performance and speed are between those of qwen-max and qwen-turbo. It is suitable for moderately complex tasks. |
qwen-max | Built-in LLM | The best-performing model in the Qwen series. It is suitable for complex, multi-step tasks. |
qwen2-72b-instruct | Built-in LLM | A balanced model in the Qwen series. Its inference performance and speed are between those of qwen-max and qwen-turbo. It is suitable for moderately complex tasks. This is a stable version of the model and will not be updated. |
Built-in LLMs support only the Q&A Test operation.
External LLMs support the Details, Q&A Test, Edit, and Delete operations.
Custom LLMs support the Details, Q&A Test, Train Model, and Delete operations. (Custom LLMs are available only for Professional Edition instances.)