Model Introduction
Last updated:2026-02-05 10:11:30
Tencent Cloud Agent Development Platform (Tencent Cloud ADP) supports the following models. You can select according to your needs.
Standard Mode
In standard mode application runtime, use the following models as the main ones:
1. Reasoning Models: Used for intent recognition and primarily responsible for improving intent-understanding accuracy.
2. Generation Models: Primarily responsible for reading comprehension and generating answers.
3. Multimodal QA Model: Used for understanding uploaded images in application dialogues. You need to set a custom model that supports image understanding.
4. Multimodal Reading Comprehension Model: Parse multiple question and image segments retrieved from the knowledge library. You need to set a custom model that supports image understanding.
5. Prompt Rewriting Model: Used to address the "fragmented context" issue in multi-round dialogues, converting blurry, abbreviated, or context-dependent user inputs into complete semantic instructions that large models can accurately understand. Tencent Cloud ADP officially recommends using youtu-rewrite for prompt rewriting.
6. AI One-Click Optimization Model: Used for one-click optimization of prompt content, role-based directives, workflow description, and prompt content for each node. At the same time, it is also used for AI code generation in workflow code nodes.
7. Real-Time Document Parsing Model: Used for document parsing of uploaded documents in QA. Tencent Cloud ADP officially recommends using youtu-parsing-sync for document parsing.
Reasoning Model
Tencent Cloud Agent Development Platform built-in youtu and DeepSeek models, and supports user customization. Standard Mode thinking model includes:
Model name | Context Length | Scene Description |
youtu-intent-pro | Maximum input: 8k Maximum output: 4k | Suitable for scenarios requiring simultaneous configuration of Q&A, documents and workflows. It provides better intent recognition and supports custom role instructions within the reasoning process, though this may increase latency. |
DeepSeek-V3.1 | 128k | Using a hybrid reasoning architecture, it possesses higher thinking efficiency, provides answers in less time, and has significantly enhanced performance in complex search tests requiring multi-step reasoning and multidisciplinary expert-level challenge tests. |
Custom Models | / | Tencent Cloud Agent Development Platform allows users to customize their experience by adding LLM model APIs that comply with the OpenAI protocol. Users can integrate these custom LLMs as generative models, with supported options including ChatGPT, Claude, Gemini, LlaMa, Qwen, Doubao, and Kimi. |
Generation Model
Tencent Cloud ADP has the built-in DeepSeek model and supports user customization. The model details and use cases are as follows:
Model name | Context Length | Scenario Description |
DeepSeek-V3.1 | 128k | Using a hybrid reasoning architecture, it possesses higher thinking efficiency, provides answers in less time, and has significantly enhanced performance in complex search tests requiring multi-step reasoning and multidisciplinary expert-level challenge tests. |
DeepSeek-R1-0528 | 128k | DeepSeek-R1 latest version has significantly improved in understanding intent, copywriting generation, programming skills, and logical reasoning. It can better understand constraints and internal logic in complex directives. With support for long-duration thinking, it can handle more complex and time-consuming tasks. |
Custom Models | / | Tencent Cloud Agent Development Platform allows users to customize their experience by adding LLM model APIs that comply with the OpenAI protocol. Users can integrate these custom LLMs as generative models, with supported options including ChatGPT, Claude, Gemini, LlaMa, Qwen, Doubao, and Kimi. |
Multi-Agent Mode
In Multi-Agent Mode, you can choose the model for each Agent. The model mainly impacts task plan and tool call in Agent Mode.
Agent model includes:
Model name | Context Length | Scenario Description |
DeepSeek-V3.1 | 128k | Using a hybrid reasoning architecture, it possesses higher thinking efficiency, provides answers in less time, and has significantly enhanced performance in complex search tests requiring multi-step reasoning and multidisciplinary expert-level challenge tests. |
DeepSeek-R1-0528 | 128k | DeepSeek-R1 latest version has significantly improved in understanding intent, copywriting generation, programming skills, and logical reasoning. It can better understand constraints and internal logic in complex directives. With support for long-duration thinking, it can handle more complex and time-consuming tasks. |
Custom Models | / | Tencent Cloud Agent Development Platform allows users to customize their experience by adding LLM model APIs that comply with the OpenAI protocol. Users can integrate these custom LLMs as generative models, with supported options including ChatGPT, Claude, Gemini, LlaMa, Qwen, Doubao, and Kimi. |
Was this page helpful?
You can also Contact Sales or Submit a Ticket for help.
Yes
No
Feedback