xAI launched a Grok add-in that brings email drafting and inbox assistance into Microsoft Outlook.
What changed
View all changes →Moonshot AI released Kimi K3, its next-generation agentic model for coding and tool use.
Thinking Machines launched Inkling, a model designed for collaborative work across long, complex tasks.
Ordered by the event or original publication date—not when a profile was reviewed.
Explore the AI market landscape
Research
9Early research & prototypes
World-model-based AI with persistent memory, reasoning, planning, controllability, and safety
AI systems that automate machine learning research and the process of building new AI models.
Energy-based models and formal verification for high-certainty critical systems
Frontier code models and ultra-long-context systems for software engineering and research
Post-transformer frontier models designed for continual learning and time-aware reasoning
Frontier open intelligence, agents, and reasoning systems
Frontier AI research focused singularly on safe superintelligence
Foundation models for computer use trained on large-scale video interaction data
Unified world models for scientific discovery, hypothesis generation, and collaborative intelligence
Preview
1Limited access & evaluation
Customizable frontier AI systems, multimodal research, and model tooling
Growth
4Public access & scaling
Agentic AI systems built for autonomous enterprise workflows
Multimodal and modular intelligence systems built for real-world deployment
Chinese frontier-style multimodal and agentic model development
Chinese frontier language and multimodal systems from the GLM family
Deployment
16Production & at scale
Consumer, cloud, and commerce distribution for the Qwen model stack
Foundation models embedded in AWS and Amazon product surfaces
General language models, tool use, coding, enterprise agents
Private on-device and server foundation models integrated across consumer devices
Chinese language, multimodal, reasoning, and agentic foundation models
Frontier multimodal models, research systems, scientific AI
Diffusion language models built for fast reasoning, coding, and production inference
Open-weight and proprietary multimodal systems at consumer scale
Platform-scale AI systems across copilots, cloud, and small language models
Multimodal models and AI-native consumer products
Frontier language models with a strong open and enterprise posture
Chinese frontier language and reasoning models with strong consumer distribution
Korean and multilingual foundation models, open models, search, agents, and cloud AI
General multimodal foundation models, reasoning, coding, video
Large language, multimodal, agentic, and media foundation models distributed through Tencent products
Frontier chat, reasoning, and consumer assistant models
World-model-based AI with persistent memory, reasoning, planning, controllability, and safety
AI systems that automate machine learning research and the process of building new AI models.
Energy-based models and formal verification for high-certainty critical systems
Frontier code models and ultra-long-context systems for software engineering and research
Post-transformer frontier models designed for continual learning and time-aware reasoning
Frontier open intelligence, agents, and reasoning systems
Frontier AI research focused singularly on safe superintelligence
Foundation models for computer use trained on large-scale video interaction data
Unified world models for scientific discovery, hypothesis generation, and collaborative intelligence
Need cross-category filtering or a broader scan of all 207 profiles?
Open the full directory →How to read the map
Category
Where an organization creates its clearest market value.
Stage
Its current path from research through durable deployment.
Trust
Editorial verification, sources, and review status stay visible.




























