Tencent / tencent/hy4-preview

Tencent: Hy4 Preview - access through LLMTR

Tencent Hy4 Preview is a preview-generation model with a 262,144-token context window, built for coding agents, long-document analysis and multi-step tool workflows. It reasons by default, returning the chain separately in `reasoning_content`; those tokens are counted inside `completion_tokens`. To turn reasoning off, send `reasoning_effort: "none"` or append `:none` to the model id - that is the only level with a measured effect here. It supports native tool calling, schema-enforced JSON through `json_schema`, and prompt caching at the lower cache-read rate. Two limits: `tool_choice: "required"` is ignored upstream, so LLMTR refuses it with a 400 instead of letting your agent loop miss the call; a named function choice works but returns `finish_reason: "stop"`, so branch on the presence of `tool_calls`, not on the label. The model is text only - an image sent to it is dropped without an error. Details are on the LLMTR Chat Completions and Tool Calling documentation pages.

Technical specifications

Canonical IDtencent/hy4-preview
ProviderTencent
Context window262,144 tokens
OperationsCHAT_COMPLETIONS
Modalitiestext

Pricing

An 8% platform margin applies to credit top-ups; model usage prices are not separately marked up.

OperationMetricUnitPrice
CHAT_COMPLETIONSINPUT_TEXTPER_1M_TOKENS$0.834000
CHAT_COMPLETIONSCACHE_READPER_1M_TOKENS$0.042000
CHAT_COMPLETIONSOUTPUT_TEXTPER_1M_TOKENS$2.501

Example usage

With existing OpenAI SDK flows, change only the base URL and model identifier.

curl https://llmtr.com/v1/chat/completions \
  -H "Authorization: Bearer llmtr-your_key" \
  -H "Content-Type: application/json" \
  -d '{"model":"tencent/hy4-preview","messages":[{"role":"user","content":"Hello"}]}'

Related models