AgentsModel picker
Local models you can run on your own hardware, for coding, images, video and voice. These are our editorial notes with facts taken from each model's own page. For rankings, we link to independent leaderboards.
| Model | Runs with | Hardware tier | Our notes |
|---|---|---|---|
| Qwen2.5-Coder (7B / 14B / 32B) | Ollama, LM Studio | Gaming GPU | Open coding model family offered on Ollama in sizes from 0.5B to 32B. Smaller tags fit laptops; 14B–32B want a strong GPU. |
| Qwen3-Coder | Ollama | Workstation | Newer agentic coding family. Check the tag sizes on its Ollama page against your memory before you download. |
Coding agents that can drive these local models: Aider, Cline, OpenCode, Kilo Code and Qwen Code. Prefer a cloud agent? Compare GitHub Copilot, Claude Code, OpenAI Codex, Gemini CLI, Cursor and Devin Desktop (ex-Windsurf) in local vs cloud coding.
Compare benchmark rankings (cloud and open models) on Artificial Analysis: language model leaderboards. We link to their leaderboards and don't republish their data.
Hardware tiers
A rough rule of thumb, not a guarantee. Check each model page for its stated requirements. LM Studio recommends at least 16GB of RAM and 4GB of dedicated VRAM.
| Tier | Memory | What usually fits |
|---|---|---|
| Laptop | 8–16 GB RAM / unified memory | Small models (about 3–8B), TTS, light image work |
| Gaming GPU | 12–24 GB VRAM | Mid-size coding models (about 7–32B), FLUX-class images, small video models |
| Workstation | 48 GB+ VRAM or 64 GB+ Mac | Large open models and heavier video generation |
Getting started
For language and coding models, install Ollama or LM Studio and pull a model tag that fits your memory. For images and video, ComfyUI is the common open-source workspace. Always check a model's license before business use.