Hy4 preview is Tencent latest Mixture-of-Experts flagship: 770B total parameters with 49B activated per token, 78 layers, 256 routed experts plus 1 shared expert, and a 1M-token context window. It is a text-only model built for coding agents, complex tool-use workflows, and productivity work, with a native MTP layer for speculative decoding. Reasoning is configurable at three levels (high, low, none) and defaults to high, so it can run as a deep chain-of-thought model for math, coding and analysis, or answer directly when latency matters. It supports native tool calling, structured outputs, and the full sampling parameter set; Tencent recommends temperature 0.9 with top_p 1.0. Tencent positions it for software engineering, office and data analysis, game prototyping, and scientific research, and reports an internal blind evaluation (163 experts, 203 engineering tasks) in which Hy4 preview averaged 2.99 against GLM 5.3 at 2.92 and Kimi K3 at 2.94. As a preview release Tencent notes known rough edges, including spending longer than necessary on reasoning and over-verifying its own work.
OpenAI-compatible — keep the SDK you already use
https://api.orcarouter.ai/v1import os
from openai import OpenAI
client = OpenAI(
base_url="https://api.orcarouter.ai/v1",
api_key=os.environ["ORCAROUTER_API_KEY"],
)
response = client.chat.completions.create(
model="tencent/hy4-preview",
messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)frequency_penaltyinclude_reasoningmax_completion_tokensmax_tokenspresence_penaltyreasoningreasoning_effortrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_p| Input / 1M tokens | $0.834 |
|---|---|
| Output / 1M tokens | $2.501 |
| Cache read / 1M | $0.042 |
| Currency | USD |
Estimate based on list price
Estimate only — actual token counts depend on the provider's tokenizer.
What developers are saying this week
GET /api/public/models/tencent/hy4-previewOpen @misc{orcarouter_hy4_preview,
title = {Hy4 preview API},
author = {Tencent},
year = {2026},
howpublished = {OrcaRouter},
url = {https://www.orcarouter.ai/models/tencent/hy4-preview}
}Tencent. (2026). Hy4 preview API. OrcaRouter. https://www.orcarouter.ai/models/tencent/hy4-preview