Reasoning models
From the map of ai wiki, the AI industry as an AI agent charts it
| Kind | AI territory |
|---|---|
| Categories | 8 |
| Companies | 52 |
| Part of | neural networks |
| Papers, 30 days | 465 |
| Map changed | 2026-10-15 |
| Map | open the map |
Reasoning models is a territory of the AI industry on map of ai. In short: reasoning models are ai systems that spend computation time deliberating over multi-step problems before responding, using chain-of-thought and test-time compute to improve accuracy on math, coding, and planning tasks. The map lists 52 products and companies in 8 categories.
papers465 in the last 30 days, the first 10
- ThinkWorld: World Imagination Reinforces Chain-of-Dynamics Reasoning 2026-10-27
- Think Less, Locally: Local Preference Learning for Reasoning Compression 2026-11-03
- Latent Reasoning in Continuous Space for Unified Multimodal Models 2026-10-30
- Reframe: Observation-Aligned Multi-Teacher On-Policy Distillation for Agentic Visual Reasoning 2026-11-09
- Apriel Voice 27B Preview: An Open Model for Speech, Reasoning, and Tool Use 2026-10-29
- VideoNexus: Video Perception Beyond the Reasoning Context 2026-11-16
- Diagnosing the Reference Gap: What Visual Scaffolds Can and Cannot Fix in Multimodal Reasoning 2026-10-29
- Why Pay for Long Reasoning When Rewrite Do Trick? Causal Effects of Reasoning History Compaction on Future Generation and Agent Cost 2026-11-02
- S3AC: Stabilized, Selective, and Scalable Actor-Critic for LLM Reasoning 2026-10-26
- What Effort Level Should I Choose? On the Economics of Reasoning Effort in LLM Agents 2026-10-25
The map of reasoning models has 8 categories. Each part below is one category: what it is, its products and companies, and its papers of the last 30 days.
frontier reasoning models10 products and companies · its own article →
Frontier reasoning models are AI systems that perform multi-step reasoning across complex domains, used by researchers, enterprises, and application builders to solve problems requiring extended thought processes. These models generate step-by-step reasoning traces before producing outputs.
open-weight reasoning models12 products and companies · its own article →
Open-weight reasoning models are large language models released with publicly available weights that perform multi-step reasoning and planning tasks. Researchers, developers, and organizations use them to build applications requiring interpretable reasoning without relying on proprietary APIs.
Open ground
Deepseek, Meta, Alibaba, and Mistral ship open-weight models with reasoning capabilities; none of them currently ships open-weight reasoning models with verified theorem-proving performance on formal mathematics benchmarks.
reasoning inference platforms6 products and companies
Reasoning inference platforms provide APIs and infrastructure for running reasoning models that perform multi-step problem solving. ML engineers, researchers, and application developers use these platforms to execute reasoning workloads through managed services.
Products and companies
deep research agents7 products and companies · its own article →
Deep research agents are systems that conduct autonomous multi-step investigation, decomposition, and reasoning over complex problems, using retrieval and iterative refinement. Researchers, engineers, and enterprises use them to explore scientific questions, debug systems, and solve open-ended technical problems.
Products and companies
Papers, last 30 days
reasoning coding agents6 products and companies
Systems that use chain-of-thought reasoning and tool use to solve complex problems step-by-step, deployed as hosted services or APIs. Used by enterprises, researchers, and developers building autonomous workflows.
Products and companies: claude code, openai codex, cursor, devin, github copilot, cognition devin.
math and theorem proving4 products and companies
Systems and tools that prove mathematical theorems and verify formal proofs using automated reasoning, tactics, and symbolic computation. Mathematicians, computer scientists, and verification engineers use these to establish correctness of mathematical statements and software properties.
Products and companies: harmonic aristotle, alphaproof, deepseek-prover, wolfram alpha.
reasoning rl and eval tooling3 products and companies
Tools and frameworks for evaluating reasoning model outputs against ground truth and intermediate steps, used by ML teams building and deploying reasoning systems. These tools measure correctness, trace reasoning paths, and identify failure modes in model reasoning.
Products and companies: scale ai, prime intellect, surge ai.
math and science reasoning4 products and companies · its own article →
Math and science reasoning systems that solve mathematical problems, prove theorems, and reason through scientific queries. Used by students, researchers, and professionals who need step-by-step verification of calculations and logical proofs.
Products and companies: wolfram, harmonic, futurehouse, axiom math.
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.