5 models

Best Local AI Models for Reasoning

These models are built to work through multi-step math, logic and analysis problems with explicit reasoning, all running locally.

LightMost popular

Qwen3 8B

Alibaba · Apache 2.0

One of the most-downloaded mid-size AI models on HuggingFace in 2026, with tens of millions of downloads. Qwen3 8B outperforms the previous Qwen2.5 14B on most benchmarks, supports hybrid thinking mode and tool calling, and runs on any GPU with 6 GB VRAM. Apache 2.0 license — fully commercial.

Min VRAM

5 GB

Context

128K

codingchatreasoningtool use

Run with Ollama

ollama run qwen3:8b
StandardBest small coder

Phi-4 14B

Microsoft · MIT

Microsoft's research-driven Phi-4 focuses on synthetic high-quality training data, producing a 14B model that outperforms much larger models on coding and mathematical reasoning benchmarks. MIT licensed with zero usage restrictions for commercial projects.

Min VRAM

8 GB

Context

16K

codinganalysisreasoning

Run with Ollama

ollama run phi4:14b
StandardMost private

DeepSeek R1 Distill 8B

DeepSeek · MIT

A distilled version of DeepSeek's R1 reasoning model that inherits chain-of-thought problem-solving capabilities in a compact 8B package. One of the few small models that can reliably work through multi-step math and logic problems without external tooling.

Min VRAM

5 GB

Context

128K

reasoningcodinganalysis

Run with Ollama

ollama run deepseek-r1:8b
Giant

DeepSeek V4 Flash

DeepSeek · MIT

The faster, more practical sibling of the 1.6-trillion-parameter DeepSeek V4 Pro. At 284B total it activates only ~13B parameters per token, so once loaded it generates far faster than its size suggests — but even at 4-bit it needs roughly 180 GB of memory, putting it in 256 GB+ unified-memory Mac or multi-GPU server territory. MIT licensed, with a 1M-token context window.

Min VRAM

~180 GB (Q4)

Context

1M

codinganalysisreasoningchat

Run with Ollama

ollama run deepseek-v4-flash
StandardHistoric first

OpenAI gpt-oss-20b

OpenAI · Apache 2.0

OpenAI's first commercially-licensed open-weight model — a historic release under Apache 2.0. MoE architecture keeps only 3.6B parameters active at inference, so it generates tokens at nearly 7B speed despite 20B total params. Supports function calling and web browsing. Over 10M downloads across Ollama and HuggingFace. Pull it with ollama run gpt-oss:20b.

Min VRAM

12 GB

Context

128K

chatcodingreasoningtool use

Run with Ollama

ollama run gpt-oss:20b

Want to mix in more filters?

Open the faceted model browser to combine VRAM, family, license, developer and task filters, then sort the results your way.

Related collections