Hy3

Tencent's 295B Mixture-of-Experts model with 21B active parameters, native 256K context support, and three reasoning modes for complex reasoning and agentic tasks.

262,144 context
Released July 6, 2026
Starting at $0.14/M input tokens
Starting at $0.58/M output tokens
Streaming
Tools
Reasoning
JSON Output
No ratings yetSign in to rate

Select Provider

All Providers for Hy3

LLM Gateway routes requests to the best providers that are able to handle your prompt size and parameters.

DeepInfra
Context: 262.1kQuant: fp8
Input
$0.14
/M tokens
Cache Read
$0.035
/M tokens
Output
$0.58
/M tokens
Get Started
NovitaAI
Context: 262.1k
Input
$0.14
/M tokens
Cache Read
$0.035
/M tokens
Output
$0.58
/M tokens
Get Started

Frequently asked questions

What is Hy3?

Tencent's 295B Mixture-of-Experts model with 21B active parameters, native 256K context support, and three reasoning modes for complex reasoning and agentic tasks. You can access it through LLM Gateway's OpenAI-compatible API with automatic provider routing, fallback, and cost analytics.

How much does Hy3 cost?

Pricing for Hy3 on LLM Gateway starts at $0.14 per million input tokens and $0.58 per million output tokens, depending on the provider. The pricing table above always reflects the current per-provider rates.

What is the context length of Hy3?

Hy3 supports a context window of up to 262,144 tokens on its largest provider deployment.

Which providers serve Hy3?

Hy3 is served by DeepInfra, NovitaAI through LLM Gateway. Requests are automatically routed to the best available provider, with fallback when a provider has issues.

Does Hy3 support tool calling and structured outputs?

Yes. Hy3 supports both tool (function) calling and structured JSON outputs through LLM Gateway.

When was Hy3 released?

Hy3 was released on July 6, 2026.