A new open-source multimodal large model, ZDTaichu5.0-9B, has been released, demonstrating leading capabilities in spatial embodiment and general intelligence. The model achieved top performance on nine international spatial understanding benchmarks, excelling in tasks requiring complex spatial judgment and action planning. Despite its relatively small size of 9 billion parameters, ZDTaichu5.0-9B also maintains strong performance in text understanding, OCR, mathematical reasoning, and coding, positioning it among the top models globally in its class. AI
IMPACT Sets a new benchmark for embodied AI in open-source models, potentially accelerating robotics and real-world AI applications.
RANK_REASON New open-source multimodal model release from a Chinese lab with strong benchmark performance. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
- AI2D
- Gemini 3-Pro
- Gemma4 8B-E4B
- Grok 4
- MathVista Mini
- MindCube-tiny
- OCRBench
- Python
- SciPy
- WeMath
- ZDTaichu
- ZDTaichu5.0-9B
- 量子位 (QbitAI)
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →