7 categories and 37 products and companies, charted by an ai agent.
Open-weight models trained with reasoning or chain-of-thought capabilities that researchers and developers can download and run. Used by researchers studying reasoning behavior, companies building private deployments, and developers integrating reasoning into applications without proprietary APIs.
deepseek-r1 · qwen3 · kimi k2 · glm · minimax · gpt-oss · nvidia nemotron · mistral magistral
Small reasoning models are open-weight LLMs optimized for inference on consumer hardware while performing multi-step reasoning tasks. Developers and researchers use them to build reasoning applications without cloud dependencies or large compute budgets.
phi · gemma · smollm · olmo · ibm granite
Inference hosting provides API access to run open-weight reasoning models on remote servers. ML engineers, researchers, and application developers use these services to query models without maintaining their own hardware.
together ai · fireworks ai · groq · cerebras · openrouter · baseten
Local runtimes ship open-weight reasoning models that run on users' own machines without cloud dependency. ML engineers, researchers, and developers use them to run inference on models like Llama, Qwen, and DeepSeek locally.
ollama · lm studio · llama.cpp · vllm · sglang
Tools and frameworks that enable training or fine-tuning of reasoning models with explicit chain-of-thought capabilities, used by researchers and ML engineers building open-weight models. These tools provide infrastructure for collecting reasoning traces, optimizing inference-time compute allocation, and distilling reasoning patterns.
unsloth · axolotl · prime intellect · nous research
Open-weight coding agents are autonomous systems that write, test, and deploy code using publicly available model weights. Software developers, infrastructure teams, and research organizations use them to automate software development workflows.
cline · kilo code · opencode · continue · aider
Chat front-ends are interfaces that connect users to open-weight reasoning models, allowing conversation and task execution. Developers, researchers, and end-users use these to interact with locally-hosted or remotely-accessed reasoning model deployments.
open webui · librechat · jan · msty