Daily Research Brief 2026-07-11

📊 Token usage: estimated from retrieval and writing scale.

Covers the latest AI research, open source and industry moves, updated daily.

Editor’s Note

The milestone this issue: OpenAI fully released the GPT-5.6 family (flagship Sol, balanced Terra, lightweight Luna) and shipped ChatGPT Work agents — Sol’s Ultra mode coordinates 4 parallel agents, with 54% better token efficiency on coding tasks, marking the shift from chat tool to autonomous desktop agent. Apple also sued OpenAI over allegedly stolen hardware trade secrets.

1. Latest arXiv Papers

1. Cognitive-structured Multimodal Agent for Multimodal Understanding, Generation, and Editing — https://arxiv.org/abs/2607.08497

2. iFLYTEK-Embodied-Omni: Unifying Vision, Language and Action in a Single Embodied Model — https://arxiv.org/abs/2607.02542

3. Gemma 4 Technical Report — https://arxiv.org/abs/2607.02770

4. WebSwarm: Recursive Multi-Agent Orchestration for Deep-and-Wide Web Search — https://arxiv.org/abs/2607.08662

5. FSD-VLN: Fast-Slow Dual-System Modeling for Aerial Long-Horizon Vision-Language Navigation — https://arxiv.org/abs/2607.08359

6. Multimodal Continuous Reasoning via Asymmetric Mutual Variational Learning — https://arxiv.org/abs/2607.00461

7. Prismata: Confining Cross-Site Prompt Injection in Web Agents — https://arxiv.org/abs/2607.08147

8. Remember When It Matters: Proactive Memory Agent for Long-Horizon Agents — https://arxiv.org/abs/2607.08716