XiaoZhan AIhubXiaoZhan AIhub
HomeAI ChatAgentsCreateCreative AgentModel catalogueTopicsPriceDocs
XiaoZhan AIhubXiaoZhan AIhub
HomeAI ChatAgentsCreateCreative AgentModel catalogueTopicsPriceDocs
XiaoZhan AIhubXiaoZhan AIhub
HomeAI ChatAgentsCreateCreative AgentModel catalogueTopicsPriceDocs
Model catalogueTencent Hunyuan Hy3

Tencent Hunyuan Hy3

NEW

Model ID

Put this value in the request body's `model` field — not the display name above.

Text generationToolsJSON256kOpen source

by tencent · Tencent Hunyuan Hy3 — a 295B MoE with 21B active, 256K context, built for reasoning, code, and Agent scenarios.

Try it now

Quick start

EndpointPOST /v1/chat/completions
AuthenticationAuthorization: Bearer sk-gpushare-…Get API key
Call styleSynchronous: one request returns the result
Base URLhttps://usapi.dflop.top/v1
cURL
curl https://usapi.dflop.top/v1/chat/completions \
  -H "Authorization: Bearer sk-gpushare-xxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "hy3",
    "messages": [
      {"role": "user", "content": "你可以帮我做什么"}
    ]
  }'

Pricing

Billed per token: input and output are priced separately and settled on the upstream's reported usage.

Rates
Input 0.072Credits/K·Output 0.24Credits/K
  • Input tokens that hit the prompt cache are billed at 24 Credits / 1M — cheaper than full-price input.
  • Failed requests (upstream 4xx / 5xx) are not billed.
  • All your API keys share one balance; when it runs out the API returns HTTP 402 `quota_exceeded`.

Sign in and any personal discount on your account will show up here. Sign in

Capabilities & specs

Category
Text generation
Context window
256k tokens
Default max output
33k tokens
Endpoint family
Chat Completions
Protocols
OpenAI Chat
Released
2026-08-23
License
Open source

Try it

Pick a key under "API access" and fire one real request to see the response — the estimated cost is shown before you send.

Try it in chat

Common errors

404 model_not_found

You passed the display name instead of the model ID.✗ Tencent Hunyuan Hy3→✓ hy3

→ Copy the Model ID at the top of this page

402 quota_exceeded

Insufficient account balance. All your API keys share one balance.

→ Add funds in your account

401 / 403

The API key is invalid, or it has a model allowlist that doesn't include this ID.

→ Check the key's status and allowed models

400 invalid_request

A request body field is invalid, or this model doesn't support the protocol endpoint you called.

→ Check your request body against the sample code above

Try it

关于 腾讯混元 Hy3 的常见问题

怎么调用 腾讯混元 Hy3?
请求体的 model 字段填 hy3,打到 https://usapi.dflop.top/v1/chat/completions,把 sk-gpushare- 开头的密钥放进 Authorization: Bearer 请求头。OpenAI、Anthropic、Google Gemini 三套官方 SDK 都能直接调用,只需把 base_url 指向本网关。
腾讯混元 Hy3 的 API 价格是多少?
  • ·输入:每 100 万 token 72 积分(≈ ¥1.2 / ≈ $0.178)
  • ·输出:每 100 万 token 240 积分(≈ ¥4 / ≈ $0.593)
  • ·缓存命中的输入:每 100 万 token 24 积分(≈ ¥0.4 / ≈ $0.059)
  • ·平台以积分计价,60 积分 = ¥1;每次调用先按上限预扣,返回后按实际用量结算,差额自动退回。
腾讯混元 Hy3 的上下文窗口有多大?
256,000 token,单次请求默认输出上限 32,768 token。
腾讯混元 Hy3 支持哪些能力?
工具调用、JSON 模式、开源权重。兼容协议:OpenAI Chat Completions。

输入价格约合每 100 万 token $0.178。本页的 Markdown 版本:/models/hy3.md