Hy4 from Tencent is a major new open-weight AI model released November 4, 2026. It brings native vision capabilities and computer use for the first time in the series, enabling it to process images, charts, screenshots and even control software via terminal or GUI interfaces.
Per the official research page at https://hy.tencent.ai/research/hy4 and Tencent's announcement at https://www.tencent.com/tencent-launches-hy4-joining-the-ranks-of-the-worlds-leading-open-source-ai-models/, it has scaled up training significantly, excelling in instruction following, long context, research tasks, financial analysis, and agentic workflows. It is positioned among the top open-weight models available.
The model page is live at https://qubax.ai/models/hy4.
Pricing
Prices as of November 8, 2026, source: qubax.ai/price-index.
| Model | Provider | Input ($/1M tokens) | Output ($/1M tokens) |
|---|---|---|---|
| Hy4 | Qubax | 0.62 | 1.86 |
| Hy4 | OpenRouter | 0.42 | 1.25 |
OpenRouter serves as the benchmark for comparison. Qubax provides access through an open market where compute providers compete on price for wholesale inference. Always verify live rates on the price index as they can change.
To start using Hy4 immediately with simple OpenAI-compatible API, sign up for access.
How to Call the Hy4 API in Python
Hy4 works seamlessly with the OpenAI Python SDK by pointing to the Qubax endpoint.
import openai
client = openai.OpenAI(
base_url="https://api.qubax.ai/v1",
api_key="your_qubax_api_key"
)
response = client.chat.completions.create(
model="hy4",
messages=[{"role": "user", "content": "Analyze this chart and summarize key trends."}],
temperature=0.7
)
print(response.choices[0].message.content)
See full setup in the /docs/quickstart.
When to Use Hy4
Hy4 shines in multimodal and long-horizon tasks such as document analysis with images, automated research, financial modeling, and software agent workflows.
Compared to GPT-4o (Qubax pricing $0.15 input / $0.60 output from latest data), Hy4 offers stronger vision and computer-use features for specialized tasks. It also stands out against other alternatives in agentic coding and research per recent comparisons like the weekly price index post.
Choose Hy4 when your application needs integrated vision + reasoning + tool use. For pure text, lower cost options may be preferable based on the live /price-index.
Get your API key and test Hy4 in production today at qubax.ai. No credit card required for initial exploration.
FAQ
How much does the Hy4 API cost on Qubax?
As of November 8, 2026, input is $0.62 per million tokens and output is $1.86 per million tokens. See the full table above and always check /price-index for updates.
What is the context window for Hy4?
Hy4 supports extended context for long documents and complex multi-step agent interactions. Exact limits are listed on the /models/hy4 page.
How do I use the Hy4 model with the OpenAI SDK?
Set base_url="https://api.qubax.ai/v1" and model="hy4" as shown in the code example above. Full details in /docs/quickstart.
Is Hy4 the cheapest option for vision tasks?
It depends on the exact workload. Compare live on the price index. For some text-only tasks, other models like GPT-4o may have lower prices.
Does Hy4 support function calling and tool use?
Yes, it fully supports the standard OpenAI tool calling format for building agents.
Where can I find more comparisons involving Hy4?
Check our latest AI Model Price Index Weekly and the dedicated model page.
Originally published at qubax.ai. Qubax gives you every top AI model with one API key, cheaper than OpenRouter.
Top comments (0)