Tencent Cloud Partner: 16 Models Free Until 16 November

Onysoft has signed a partnership agreement with Tencent Cloud and become a Tencent Cloud Partner. You can use the Tencent Cloud versions of Hy, DeepSeek, GLM, Kimi and MiniMax models (16 model IDs) free until 16 November 2026, 23:59 (UTC+3) with your existing Onysoft API key: add :tencent to the end of the model ID, and these requests are not deducted from your balance. Onysoft also serves the paid Tencent Hy4, Hy4 preview and Hy3 models through a direct integration with Tencent Cloud.

During the free period, a daily usage limit per account and a limit of 64K input / 16K output tokens per request apply; only text and tool calls (function calling) are supported. Free models are open to verified accounts: you must have signed in with Google or GitHub, or topped up your balance at least once with a real payment. These requests are processed in Tencent Cloud's Singapore region. For DeepSeek V4.1 Flash, Tencent Cloud forwards the request to the infrastructure of the model developer (DeepSeek). Total capacity is limited; if it runs out, free usage may stop for the rest of the day or until the end of the period.

This article covers the free models and their limits, the scope of the partnership, which Hy model fits which job, how to send your first request, and frequently asked questions. The current list of all models is in the model catalog; parameters and error codes are in the API documentation.

At a Glance: The Partnership and the Free Period

Topic	Details
Partnership	Onysoft is a Tencent Cloud Partner
Free period	16 models, free until 16 November 2026, 23:59 (UTC+3) (full list below)
Free usage	Add :tencent to the end of the model ID; requests are not deducted from your balance
Free period limits	Daily usage limit per account; 64K input / 16K output tokens per request; text and tool calls
Where the free models are processed	Tencent Cloud's Singapore region. For DeepSeek V4.1 Flash, Tencent Cloud forwards the request to the infrastructure of the model developer (DeepSeek).
Capacity	Total capacity is limited; if it runs out, free usage may stop for the rest of the day or until the end of the period
Direct integration (paid)	Onysoft serves Hy4, Hy4 preview and Hy3 through a direct integration with Tencent Cloud
Onysoft model IDs	tencent/hy4, tencent/hy4-preview, tencent/hy3
Endpoint	https://api.onysoft.com/v1 (OpenAI-compatible)
Authentication	Onysoft API key (sk-ony- prefix)
Tencent Cloud account	Not required

You reach the covered models with the same API key as every other model in the catalog. Paid requests are deducted from the same balance and appear in the same usage records; free models (IDs ending in :tencent) are not deducted from your balance during the period. To switch from one model to another, change the model field in your request.

Models Free Until 16 November

The 16 model IDs below are free until 16 November 2026, 23:59 (UTC+3). You do not need a new key: use your existing API key and add :tencent to the end of the model ID. These requests are not deducted from your balance.

Model family	Model ID
Tencent Hy	tencent/hy4:tencent
Tencent Hy	tencent/hy4-preview:tencent
Tencent Hy	tencent/hy3:tencent
DeepSeek	deepseek/deepseek-v4.1-flash:tencent
DeepSeek	deepseek/deepseek-v4-flash-202605:tencent
DeepSeek	deepseek/deepseek-v4-pro-202606:tencent
DeepSeek	deepseek/deepseek-v4-pro-0813:tencent
GLM	z-ai/glm-5.3:tencent
GLM	z-ai/glm-5.3-flash:tencent
GLM	z-ai/glm-5.3-flashx:tencent
GLM	z-ai/glm-5.2:tencent
Kimi	moonshotai/kimi-k3:tencent
Kimi	moonshotai/kimi-k2.7-code:tencent
Kimi	moonshotai/kimi-k2.7-code-highspeed:tencent
MiniMax	minimax/minimax-m3:tencent
MiniMax	minimax/minimax-m3.1-flash:tencent

Example: a curl request to tencent/hy4:tencent. To try another model, replace the model field with one of the IDs in the table.

curl https://api.onysoft.com/v1/chat/completions \
  -H "Authorization: Bearer sk-ony-YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "tencent/hy4:tencent",
    "max_tokens": 8192,
    "messages": [{"role": "user", "content": "What is Tencent Cloud? Explain in two sentences."}]
  }'

• Usage limit: A daily usage limit applies per account.
• Request size: 64K input / 16K output tokens per request.
• Request type: Only text and tool calls (function calling) are supported.
• Eligibility: Free models are open to verified accounts: you must have signed in with Google or GitHub, or topped up your balance at least once with a real payment (card or bank transfer).
• Region: Requests are processed in Tencent Cloud's Singapore region. For DeepSeek V4.1 Flash, Tencent Cloud forwards the request to the infrastructure of the model developer (DeepSeek). (Model ID: deepseek/deepseek-v4.1-flash:tencent.)
• Capacity: Total capacity is limited; if it runs out, free usage may stop for the rest of the day or until the end of the period.

The free period ends on 16 November 2026 at 23:59 (UTC+3); after that, the :tencent versions are closed. You can keep using the paid tencent/hy4, tencent/hy4-preview and tencent/hy3 IDs. Current limits and error codes are in the free models section of the documentation.

Tencent Cloud and the Tencent Hy Models

Tencent Cloud is Tencent's cloud computing platform. Tencent Cloud's large language model service offers Tencent's own Hy (Hunyuan) models through a unified API. Onysoft serves Tencent Hy4, Hy4 preview and Hy3 through a direct integration with Tencent Cloud; you reach these models with your Onysoft API key, through the same endpoint as the other models in the catalog.

Tencent released Hy4 on 4 November 2026. Hy4 preview is the preview release, with the same context and output limits as Hy4. Hy3, with a 256K-token context window and 128K tokens of max output, is a separate option for jobs that generate long text or code; the values are in the table in the next section.

Which Model for Which Job?

Model	Onysoft ID	Context	Max input	Max output
Hy4	tencent/hy4	1M	960K	64K
Hy4 preview	tencent/hy4-preview	1M	960K	64K
Hy3	tencent/hy3	256K	192K	128K

Source: Tencent Cloud model list and Hy API guide (updated 4 November 2026). Context, input and output values are in tokens. For the :tencent versions in the free period, a limit of 64K input / 16K output tokens per request applies.

• Hy4 (tencent/hy4): For work that needs a large context. Its 1M context window leaves room to process large codebases and long documents in a single request.
• Hy4 preview (tencent/hy4-preview): The preview release, with the same context and output limits as Hy4; to compare the two on the same prompt, change only the model field.
• Hy3 (tencent/hy3): For jobs that generate long responses. Its 128K max output is higher than Hy4's 64K limit, which matters for jobs that generate long reports or large code files.

What Developers Get

• Free until 16 November: You can try the :tencent versions of Hy4, Hy4 preview, Hy3 and the other models in the table without any deduction from your balance until the period ends.
• Tencent Cloud partnership: Onysoft serves the paid Hy4, Hy4 preview and Hy3 models through a direct integration with Tencent Cloud.
• One key, one endpoint: You do not need to open a Tencent Cloud account or manage a separate API key; you reach all 783 models in the catalog with the same Onysoft key (free models require account verification).
• Your existing code works: Because the endpoint follows the OpenAI schema, the official OpenAI SDKs and tools that use this schema work by changing only base_url, the API key and the model name.
• One dashboard: You track your Hy4, Hy4 preview and Hy3 usage alongside other models in the usage records of your Onysoft dashboard.
• Easy to try: You can try the same prompt on Hy4, Hy3 and other catalog models by changing only the model field, or use the Playground to try them without writing code.

How to Use It: Your First Request

To get started, create an account and generate an API key with the sk-ony- prefix in your dashboard. Add balance for paid models; the conditions for the :tencent versions in the free period are in the section above. The curl example below sends a non-streaming request to Hy3; the text of the reply is in choices[0].message.content:

curl https://api.onysoft.com/v1/chat/completions \
  -H "Authorization: Bearer sk-ony-YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "tencent/hy3",
    "max_tokens": 16384,
    "messages": [{"role": "user", "content": "Summarise this customer email in three bullet points: ..."}]
  }'

A streaming request to Hy4 with the official openai Python package:

from openai import OpenAI

client = OpenAI(
    base_url="https://api.onysoft.com/v1",
    api_key="sk-ony-YOUR_KEY",
)

stream = client.chat.completions.create(
    model="tencent/hy4",
    messages=[
        {"role": "system", "content": "Answer briefly and clearly."},
        {"role": "user", "content": "Write an error-handling checklist for a REST API."}
    ],
    stream=True,
)
for chunk in stream:
    if chunk.choices and chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

For Hy4 preview, set the model name to tencent/hy4-preview. See the API documentation for supported parameters and error codes, and the model catalog for the current list of models you can reach with the same key. Details on moving existing OpenAI code are in the OpenAI SDK base_url guide.