Introducing OpenReward | General Reasoning

General Reasoning is a long horizon capabilities company.

Introducing OpenReward | General Reasoning Release March 2026 Introducing OpenReward An open resource for serving RL environments Reinforcement learning (RL) is a data-hungry paradigm. Unlike pre-training the bulk of data is not available on the internet. Instead, data is generated from environments sold by private data vendors. Because environments are excludable, and vendor contracts increasingly exclusive, the status quo favours incumbents and the closed ecosystem. Open models are therefore struggling to stay competitive in the reinforcement learning era. As part of our commitment to open r

saved by

• Tom Zheng

related reading

• Training a Misaligned Reward Seekeralignment.anthropic.com
• Reward Hacking in Reinforcement Learning | Lil'Loglilianweng.github.io
• RL Environments and RL for Science: Data Foundries and Multi-Agent Architecturesnewsletter.semianalysis.com
• State of RL for reasoning LLMs | A. Weersaweers.de
• Introducing Marin: An Open Lab for Building Foundation Models | Marinmarin.community
• Measuring Reward-Seeking by Instilling Contrastive Beliefsalignment.openai.com
• A Taxonomy of RL Environments for LLM Agentsleehanchung.github.io
• Systematic Reward Hacking and Prime Sprintsprimeintellect.ai
• Keep the Tokens Flowing: Lessons from 16 Open-Source RL Librarieshuggingface.co
• GitHub - open-compass/opencompass: OpenCompass is an LLM evaluation platform, supporting a wide range of models (Llama3, Mistral, InternLM2,GPT-4,LLaMa2, Qwen,GLM, Claude, etc) over 100+ datasets.github.com
• [2607.18966] Measuring Reward-Seeking via Contrastive Belief Updatesarxiv.org
• Scaling Agentic RL: 365,000+ Environments for SWE, Terminal, and Searchprimeintellect.ai