Standard & Paid Price
| Meter | Standard Price |
|---|---|
| Input price | $0.8340 / 1M tokens |
| Completion price | $2.5010 / 1M tokens |
| Cache read price | $0.0420 / 1M tokens |
Description
Tencent Hunyuan 4 preview (Hy4 preview) — 1M-token context window (960K max input, 64K max output) with deep thinking (which Tencent labels preserved thinking), structured output, function calling and prompt caching. Text-only. Preview SKU: Tencent retires a preview model once its general-availability version ships.
Capabilities
- Tasks
- chat, reasoning
Decision evidence
Hy4 preview is Tencent Hunyuan's open-weight mixture-of-experts model (770B total, 49B active), released on 2026-08-28 under Apache-2.0, with a 1M-token context window, up to 960K input and 64K output on Tencent Cloud, and text-only input; thinking is on by default. Tencent lists overlong reasoning and over-verification as known issues; in the reviewed tests it drew better shadows than DeepSeek V4.1 Flash on one scene and scored 48/60 in WorkBuddy but 43 in Claude Code on kzhu's test.
2026-09-25 · Markdown evidence · JSON data
API endpoints
POST /v1/messages(anthropic)POST /v1/chat/completions(openai)