Tencent Hunyuan Commands & Prompt Reference

22 commands & parameters · 4 categories · Source: Tencent Cloud Hunyuan ChatCompletions API

Tencent Hunyuan's ChatCompletions endpoint follows Tencent Cloud's API conventions rather than the OpenAI shape, so parameter names are PascalCase and three common parameters (Action, Version, Region) travel with every request. The conversation itself goes in Messages.N, capped at 40 entries ordered oldest to newest, with system first if present and user and assistant alternating.

What distinguishes Hunyuan from most chat APIs is the size of its search and enhancement surface. EnableEnhancement turns on capability augmentation, ForceSearchEnhancement makes it mandatory, EnableSpeedSearch trades depth for a faster first token, and SearchInfo plus Citation control what search metadata comes back. Several parameters are model-gated: tool calling works on hunyuan-turbos, hunyuan-t1 and hunyuan-functioncall, while EnableThinking currently applies only to hunyuan-a13b.

All Commands

Core Request Parameters (8)

Command / SyntaxWhat it doesNotes
ModelModel name to callRequired. Example: hunyuan-turbos-latest. Billing differs by model.
Messages.NChat context array, ordered oldest to newestRequired. Up to 40 entries. Role values: system, user, assistant, tool. System, if present, must be first; user and assistant alternate.
StreamStreaming response switchDefaults to false. When true, results arrive incrementally over SSE in Choices[n].Delta.
StreamModerationContent moderation switch for streaming outputBoolean.
TemperatureSampling temperature controlling output randomnessFloat. Set either Temperature or TopP, not both.
TopPNucleus sampling probability thresholdFloat. Set either Temperature or TopP, not both.
SeedRandom seed for reproducible outputNon-zero positive integer up to 10000. Not recommended unless needed; unsuitable values degrade quality.
Stop.NCustom stop strings that end generationArray of strings.

Tools & Reasoning Parameters (3)

Command / SyntaxWhat it doesNotes
Tools.NList of callable tools for function callingEffective only on hunyuan-turbos, hunyuan-t1 and hunyuan-functioncall models.
ToolChoiceTool usage modeValues: none, auto, custom. Defaults to auto. Effective on hunyuan-turbo and hunyuan-functioncall models.
EnableThinkingChain-of-thought reasoning switchEnabled by default. Currently effective only on the hunyuan-a13b model.

Search & Enhancement Switches (8)

Command / SyntaxWhat it doesNotes
EnableEnhancementCapability enhancement switch, such as searchBoolean.
ForceSearchEnhancementForce search enhancement onBoolean.
EnableSpeedSearchTurn on the fast search variantDefaults to false. When on and search is triggered, streaming returns the first token sooner.
SearchInfoReturn SearchInfo in the response when search is triggeredDefaults to false.
CitationSearch citation superscript switchBoolean.
EnableMultimediaMultimedia output switchBoolean.
EnableRecommendedQuestionsRecommended follow-up questions switchBoolean.
TopicChoiceTopic passed in by the callerString.

Common API Parameters (3)

Command / SyntaxWhat it doesNotes
ActionCommon parameter; the API action nameFixed value for this endpoint: ChatCompletions. Required.
VersionCommon parameter; the API versionFixed value for this endpoint: 2023-09-01. Required.
RegionCommon parameter; the regionOptional for this endpoint.

Frequently Asked Questions

Why do Hunyuan parameters look different from OpenAI's?

The endpoint follows Tencent Cloud API conventions: PascalCase names, and every request carries the common Action and Version parameters (ChatCompletions and 2023-09-01 for this endpoint).

How long can the conversation context be?

Messages.N holds up to 40 entries, ordered oldest to newest. If total Content exceeds the model's input limit, the earliest content is truncated and the tail is kept.

Which models support tool calling and thinking?

Tools and ToolChoice are effective on hunyuan-turbos, hunyuan-t1 and hunyuan-functioncall. EnableThinking currently applies only to hunyuan-a13b.

Every entry on this page is a literal, typeable token taken from the official documentation linked above. Command sets in AI tools change quickly - the in-product command menu is always authoritative for your installed version.

Related prompt references

DeepSeek (22) · GLM (Z.ai) (23) · Qwen (Alibaba Model Studio) (29) · All AI prompt references