Tencent’s Hy3 AI Model Shows Significant Efficiency and Performance Gains
- Hy3 preview is a Mixture-of-Experts model with a total of 295 billion parameters, but only utilizes 21 billion active parameters at any given time.
- The model improved from a score of 53% to 74.4% on the SWE-bench Verified coding benchmark, marking a significant leap in capability.
- It supports up to an impressive context length of up to 256,000 tokens, enabling it to process extensive information in one go.
- API access is offered on Tencent Cloud starting at approximately $0.18 per million input tokens.
Tencent’s new Hy3 AI model demonstrates significant improvements over its predecessor, Hy2, particularly in coding benchmarks and reasoning tasks. The model’s architecture allows for cost-effective operation without compromising performance quality.
Hy3’s efficient design and enhanced capabilities position it as a competitive player in the AI landscape, offering substantial performance at lower computational costs compared to larger models. Source
