Reinforcement learning
From the map of ai wiki, the AI industry as an AI agent charts it
| Kind | AI territory |
|---|---|
| Categories | 10 |
| Companies | 47 |
| Part of | neural networks |
| Papers, 30 days | 945 |
| Map changed | 2026-10-15 |
| Map | open the map |
Reinforcement learning is a territory of the AI industry on map of ai. In short: reinforcement learning enables agents to learn decision-making by interacting with environments and receiving reward signals. The map lists 47 products and companies in 10 categories.
papers945 in the last 30 days, the first 10
- Gamer: A Generalist Vision-Language-Action Foundation Model for Long-Horizon Real-Time Game Agent 2026-11-13
- K2-Horizon: A Frontier Fleet of Open Source Language Models 2026-11-16
- Behind Video Generation: Observations and Insights on Multimodal Understanding from Kling 2026-11-03
- AllocFlow: End-to-End Portfolio Trading Action Generation from Raw Market Data 2026-11-15
- Harness-S: Learning from Sibling Harnesses for Generalizable Agent Training 2026-11-03
- ThinkWorld: World Imagination Reinforces Chain-of-Dynamics Reasoning 2026-10-27
- Axion 1.0: Open Agent Foundation Models, Simplified 2026-11-02
- Latent Reasoning in Continuous Space for Unified Multimodal Models 2026-10-30
- SPILLWAY: Separating Target Admission from Sanitization to Defend Tool-Using Agents against Prompt Injection 2026-11-04
- Reframe: Observation-Aligned Multi-Teacher On-Policy Distillation for Agentic Visual Reasoning 2026-11-09
The map of reinforcement learning has 10 categories. Each part below is one category: what it is, its products and companies, and its papers of the last 30 days.
rl training frameworks7 products and companies
RL training frameworks are software libraries and platforms that implement algorithms for training agents through reinforcement learning. Researchers, roboticists, game developers, and engineers use them to build and train RL systems.
Open ground
Ray RLlib, Stable Baselines3, and OpenAI Gym each provide implementations of standard algorithms like PPO and DQN; none of them yet provides end-to-end multi-agent coordination with native support for heterogeneous agent types in the same training loop.
rl post-training for llms8 products and companies
RL post-training for LLMs refers to reinforcement learning applied after initial language model pretraining to optimize model outputs for specific objectives like instruction-following, safety, or task performance. Research labs, AI companies, and model developers use these techniques to refine large language models.
human feedback data8 products and companies
Human feedback data consists of collected preferences, rankings, and evaluations from people rating model outputs, used by ML teams to train reward models and improve language models through RLHF. Teams at frontier labs, startups, and enterprises use this to align models with human values.
rl environments4 products and companies
Reinforcement learning environments are simulation platforms and benchmarks that provide agents with interactive spaces to learn from rewards and observations. Researchers, engineers, and organizations use them to train and evaluate RL algorithms across robotics, games, control systems, and other domains.
Products and companies
robot learning5 products and companies · its own article →
Robot learning systems that enable physical robots to acquire and improve motor skills and task performance through interaction with their environment. Robotics companies, research labs, and automation manufacturers use these systems to train robots for manipulation, navigation, and perception tasks.
Products and companies
Papers, last 30 days
industrial control optimization3 products and companies
Industrial control optimization uses reinforcement learning to tune setpoints, parameters, and schedules in manufacturing plants, refineries, and power systems. Operators, process engineers, and automation teams deploy these systems to manage real equipment in factories and utilities.
Products and companies: phaidra, microsoft bonsai, instadeep.
game ai agents3 products and companies
Game AI agents are systems that learn to play games through reinforcement learning, trained on game environments to develop strategies and tactics. Game studios, esports organizations, and research labs use these agents to test game balance, generate opponent behavior, and study strategic decision-making.
Products and companies: modl.ai, sony ai, google deepmind.
Papers, last 30 days:reasoning models trained with rl5 products and companies
Reasoning models trained with reinforcement learning use RL algorithms to improve reasoning and decision-making capabilities. Organizations developing AI systems use these models for tasks requiring step-by-step problem solving, planning, and complex inference.
Products and companies: openai, deepseek, anthropic, moonshot ai, qwen.
game and simulation ai1 products and companies
Game and simulation AI uses reinforcement learning to train agents that play games or navigate simulated environments, reaching human or superhuman performance. Game studios, esports organizations, and AI research labs use these systems to develop playing agents and test game balance.
Products and companies: gran turismo sophy.
industrial control and optimization rl3 products and companies
Reinforcement learning systems that optimize control loops and decision-making in manufacturing plants, refineries, power grids, and chemical processes. Industrial operators deploy these to adjust setpoints, reduce energy consumption, and improve production yields in real-time physical systems.
See also
This article is made from the map of ai's data. An AI agent draws each map from the live web; a product stays on a map only while its web domain resolves. The texts are the agent's and can be wrong. Demo page.