Hunyuan Hy3
Tencent’s open flagship, switching between quick answers and deep reasoning.
About Hunyuan Hy3
Hy3 is the model behind Tencent’s own products — its coding assistant, the Yuanbao chatbot and features inside WeChat. It is a mixture-of-experts model with 295 billion parameters, 21 billion active per token, and Tencent says it reasons as well as flagships two to five times its size. Its model card reports 90.4% on GPQA Diamond and 78% on SWE-bench Verified.
It has two speeds: a no-think mode for direct answers and a high-effort mode for hard problems. Tencent says it cut its rate of made-up answers to 5.4%. The weights are open under Apache 2.0.
Where it’s strong
Coding agents on a budget.
Office and productivity assistants.
Mixed work that switches between quick answers and deep reasoning.
Where to be careful
Text only.
Tencent’s own figures still show a 12.7% error rate on commonsense questions, and its benchmarks are self-reported.
Price
For scale: 1,000 requests of about 3,000 tokens in and 1,000 out cost $0.97 — a long prompt and a page of answer each. Every rate is published and stays the same from one request to the next. See every model.
Specs
hy3Call it
Use the OpenAI SDK you already have — point it at Roar AI and name the model. Read the docs.
from openai import OpenAI
client = OpenAI(base_url="https://api.roar-ai.com/v1", api_key="roar_live_…")
reply = client.chat.completions.create(
model="hy3",
messages=[{"role": "user", "content": "Summarise this contract in plain English."}],
)Compare Hunyuan Hy3
Questions
How much does Hunyuan Hy3 cost?
Input is $0.1365 per 1M tokens; output is $0.5655 per 1M tokens. For scale, 1,000 requests of about 3,000 tokens in and 1,000 out cost $0.97. The rate is published and does not change from one request to the next.
Can I use Hunyuan Hy3 alongside other models?
Yes. Every model on Roar AI is on the same API key and the same monthly invoice, so switching from Hunyuan Hy3 to another model is a change to one word in your code.
Is Hunyuan Hy3 open source?
Its weights are open under the Apache 2.0 licence, so you could run it yourself. Through Roar AI you call it like any other model, with no servers to manage.