SenseNova-MARS: Open Source AI Agent Outperforms Gemini-3-Pro in Multimodal Tasks
Analysis
SenseNova-MARS, a new Open Source agent, is making waves by surpassing Gemini-3-Pro in key benchmark tests, demonstrating impressive capabilities in multimodal search and reasoning. This groundbreaking model, developed by SenseTime, opens new avenues for developers and users to explore advanced AI functionalities.
Key Takeaways
- •SenseNova-MARS is an Open Source multimodal AI agent.
- •It outperforms Gemini-3-Pro in key benchmarks.
- •The model supports dynamic visual reasoning and deep integration of text and image search, facilitating complex tasks.
Reference / Citation
View Original"SenseNova-MARS is the first Agentic VLM model supporting dynamic visual reasoning and deep integration of text and image search; it can plan steps and call tools itself, easily handling various complex tasks, allowing AI to truly have "execution capabilities.""
雷
雷锋网Jan 30, 2026 02:45
* Cited for critical analysis under Article 32.