Archive — CN — technode
Tencent upgrades Yuanbao AI with self-developed model Hunyuan T1
Tencent upgrades its AI chatbot Yuanbao by introducing two enhanced models: Hunyuan T1 and DeepSeek V3-0324, improving capabilities in code generation and reasoning. Hunyuan T1 scores 87.2 on the MMLU benchmark, competitive yet below OpenAI's leading model. The pricing strategy is aligned with competitors, making it accessible.
Key points
- Tencent announces upgrades to Yuanbao AI with new models Hunyuan T1 and DeepSeek V3-0324.
- Hunyuan T1 scores 87.2 on the MMLU benchmark, demonstrating advanced capabilities.
- It surpasses Tencent's previous model DeepSeek-R1 (84) but is less effective than OpenAI's Qwen-1 (89.3).
- The pricing structure for Hunyuan T1 is competitive, charging RMB 1 ($0.14) per million input tokens.
Why it matters
- The upgrades reflect the ongoing trend in 2025 of enterprises enhancing AI solutions for broader applications.
- The competition in the AI market is increasing, reminiscent of the rapid developments seen in AI platforms in 2024.
Tencent's launch of Hunyuan T1 showcases its commitment to competing in the AI sector, enhancing Yuanbao's functionalities while challenging established rivals with competitive pricing.