DeepSeek · KV Cache Compression, MoE, Long-Context Agents
Microsoft (MSR Montreal) · Coding Agents, Benchmark, Feature Reconstruction
Google (Research) · Social Reasoning, Multi-Agent Simulation, LLM Evaluation
NVIDIA · Full-Duplex Speech, Streaming ASR, Conversational AI
Ant Group · Payment-Domain LLM, Benchmark, Robustness
LimiX · Tabular Foundation Model, Causal Discovery, In-Context Learning
Zing Team · World Model, Playable Generation, Real-Time Interaction
Salesforce · Enterprise LLM, Agentic Tool Use, GRPO
DeepCybo · Physical Foundation Model, Embodied AI, Next-Token Prediction
Atria · Agentic Foundation Model, Human-AI Collaboration, Scientific Research