Hunyuan Hy3

Tencent
hy3
Released Jul 6, 2026
Providers
2
Context
262K
Input
$0.13/M
Output
$0.53/M

Tencent Hunyuan's 295B/21B MoE model for real-world business scenarios: native 256K context, three reasoning modes, strong coding, long-form comprehension, multi-turn dialogue, and agentic task execution.

deepinfra/hy3Quantfp4
Input
$0.13/M
Output
$0.53/M
Context
262K
Max out
131K
Cache read
$0.03/M
Cache write
$0.13/M
Thinking out
$0.53/M
novita/hy3QuantUnspecified
Input
$0.14/M
Output
$0.58/M
Context
262K
Max out
66K
Cache read
$0.04/M
Cache write
$0.14/M
Thinking out
$0.58/M