HugstonOne

A Workflow/Software for AI inference.

Double Memory Layer Image to text / Text to text Offline/Online Mode for model and data isolation Load LLM Models in CLI or SERVER (both with image processing mode integrated Local API with CORS option GPU (Nvidia) / CPU support Instant Preview of the code Code Editior also with Preview GGUF support 10000+ LLM Models Session and minisessions with persistent memory Offline/Online Button for completely offline (cutting all connections) Terminal for advanced use (with edit mode active) Upload files (only TXT and JPEG/PNG) for now Export tables in CSV and save sessions/tabs/preview Easy to create Webpages/Graphic interfaces Great for creating games, tables, maps with data, charts, plots Good coding assistant (hopping from tab to tab, conserving memory, solving the ctx size and memory issue) Live news feed on the app while working (can retrieve all news/radio/podcast feeds) Ideal for Medical Research (with free medical models available from HuggingFace or Hugston.com) AI-powered diagnostics for X-rays, MRI, drawings, 3D modeling, etc. (via Image-to-Text) Multitasking like a browser/app, now equipped with AI—no subscription or worry about costs

Our Strongest Points: We don’t collect user data No forced updates included in the app No telemetry No HTTP involved if (load with local CLI) No internet required unless explicitly chosen by the user No API key required Supports long lines of code Input/Output (50000+ tokens tested/tab) No refusal Using the right LLM model (potential for 200,000+ code lines, divided into 5000-20000 lines at once) Completely free

Models supported: Supported Models: Qwen Qwen2: 1.8B, 3B, 7B, 32B Qwen2.5: 325B (Max) Qwen3: 4B, 8B, 14B, 30B, 480B, 80B, 235B, 46B, 14B (via A3B/Thinking), 32B (QwQ), Qwen3-Next, Qwen3-Coder, Qwen3-VL-Series etc.

Supported Models: GLM GLM-4.5: 355B GLM-4.6: 355B (same base) GLM-4-0414: 32B Supported Models: Kimi Kimi K2: 1T Kimi k1.5: 500B

Supported Models: DeepSeek DeepSeek-V3.1-Terminus: 685B DeepSeek-R1: 685B DeepSeek-TNG-R1: 685B All DeepSeek lines are under the "DeepSeek" series (685B, 27B, 400M, etc.)

Supported Models: Mistral

Mistral Magistral Medium: 7B-400B

Supported Models: Falcon (TII) Falcon-H1: 34B Falcon-Edge: 3B Falcon-1: 34B

Supported Models: Gemini / Google DeepMind Gemma 3: 27B Gemma 3n: 2B-4B Gemma 3 270M: 270M

Google DeepMind models under "Gemini/Gemma" series

Supported Models: Llama / Meta Llama 4 Behemoth: 2T Llama 4 Maverick: 400B Llama 3.1-Tulu-3: 405B Llama 3.1-Nemotron-Ultra: 253B Llama-3.3-Nemotron-Super: 49B Llama 3.1: 32B (implied) All Meta AI models under "Llama 3.1/4" series

Supported Models: OpenAI GPT-oss-120b: 120B GPT-oss-20b: 20B

Supported Models: Baidu Baidu ERNIE series, mainly 1B and 28B

Supported Models: Kuaishou KAT-V1-200B: 200B KAT-V1-40B: 40B

Supported Models: Tencent Hunyuan Hunyuan-A13B: 80B

Supported Models: Xiaomi MiMo-7B-RL-0530: 7B MiniMax-M1-M2: 456B

All IBM models And many more...