6 categories and 33 products and companies, charted by an ai agent.
Frontier reasoning models are large language models trained to perform extended reasoning and planning across complex problems, used by researchers, enterprises, and developers building systems that require step-by-step problem solving.
openai o-series and gpt-5 · claude · gemini · grok · deepseek · qwen · kimi · glm · minimax
Open-weight reasoning models are neural networks with published weights that perform multi-step reasoning and generate intermediate thinking steps before producing answers. Researchers, companies, and developers use them to build applications that require transparent reasoning chains and local deployment.
mistral magistral · nvidia nemotron · gpt-oss · phi · olmo
Reasoning inference and serving encompasses systems that execute complex multi-step reasoning tasks and deploy reasoning models in production environments. Data scientists, researchers, and enterprises use these systems to run inference on reasoning models like o1 and similar architectures that perform extended chain-of-thought computations.
together ai · fireworks ai · groq · cerebras · openrouter
Deep research agents are systems that perform multi-step reasoning and investigation to answer complex questions, used by researchers, analysts, and knowledge workers who need systematic exploration of problems. They decompose queries into sub-questions, gather information iteratively, and synthesize findings across sources.
perplexity · chatgpt deep research · manus · genspark · you.com
Systems that combine language models with step-by-step reasoning to solve complex problems by generating intermediate reasoning steps before producing outputs. Used by researchers, enterprise teams, and application developers building AI agents that need to work through multi-step tasks.
claude code · codex · cursor · cognition devin · windsurf · github copilot
Math and science reasoning models are AI systems trained to solve mathematical proofs, physics problems, chemistry equations, and scientific derivations through step-by-step logical inference. Researchers, engineers, and students use them to verify solutions and explore problem spaces.
harmonic · futurehouse · sakana ai