Hunyuan
Hunyuan spans text, image, video and 3D generation, with several components open-sourced.
Full Hunyuan record →Head to head
Hunyuan (Tencent) and ERNIE Bot (Baidu) both sit in Writing. ERNIE Bot shipped more tracked changes in the last 30 days (15 vs 12). Every value below comes from the latest crawl of the vendors' own pages.
| Attribute | Hunyuan Tencent | ERNIE Bot Baidu |
|---|---|---|
| Segment | Writing | Writing |
| Version | Hunyuan | 5.0 |
| Entry cost | Free trial quota of 1M tokens. Pay-as-you-go pricing applies afterward. | Pay-as-you-go API pricing with select free trial skills/components. Ernie 5.0 at 0.006-0.01 RMB per 1k input tokens. |
| Pricing tiers | Free Trial Package free | Pay-as-you-go API free |
| Model stack | Hunyuan-a13b · Hunyuan-embedding · Hunyuan-role-latest · Hunyuan-t1-vision · Hunyuan-T2 · Hunyuan-translation · Hunyuan-translation-lite · Hunyuan-turbos-vision · Hunyuan-turbos-vision-video · Hunyuan3D · HunyuanVideo · Tencent HY Vision 1.5 Instruct | DeepSeek-V3.1 · ERNIE 4.5 (open weights) · ERNIE 4.5 Turbo · ERNIE 4.5 Turbo VL · ERNIE 4.5 VL 28B A3B · ERNIE 5.0 · ERNIE X1.1 Preview · ernie-4.5-turbo-128k · ERNIE-5.0-Thinking-Preview |
| Context window | 256K tokens | 119k |
| Public API | ||
| Routes models | ||
| Changes / 30d | 12 | 15 |
| Origin | China | China |
| Capabilities | Image generation, Video generation, 3D asset generation, WeChat integration, Text Generation, Roleplay, Translation, Multimodal Vision | Multimodal input, Search integration, Digital humans, Enterprise Qianfan platform, Multi-Agent, RAG, Multimodal, OCR |
Hunyuan spans text, image, video and 3D generation, with several components open-sourced.
Full Hunyuan record →Baidu's assistant (Wenxin Yiyan), backed by the ERNIE model line and integrated into Baidu Search.
Full ERNIE Bot record →