README.md

---
title: README
sdk: static
pinned: false
---

[![DecisionNode](https://huggingface.co/spaces/bynn-intelligence-com/README/resolve/main/decisionnode-banner.png)](https://decisionnode.com)

# Bynn Intelligence

Bynn Intelligence builds **DecisionNode**, a family of System One decision models for software. DecisionNode models read text and images and return typed, calibrated decisions (a choice, a score, a yes-or-no probability or a number) instead of free text, so applications can act on every answer directly.

A System One model is a class of AI model that returns typed, probabilistic decisions from a fixed answer space instead of generating text. Given a piece of state and a set of questions defined in code, it answers each one with a schema-valid value such as a chosen option, a score, or a probability, and then stops.

## Models

| Model | Availability | Built for |
|---|---|---|
| **[DecisionNode Flash](https://decisionnode.com/docs/models/flash)** | [Hosted API](https://decisionnode.com/docs/api) | speed and accuracy together: decisions in milliseconds, at the lowest hosted cost |
| **[DecisionNode](https://decisionnode.com/docs/models/decisionnode)** | [Hosted API](https://decisionnode.com/docs/api) | the highest accuracy, with boosters, knowledge, video and fine-tuning |
| **[DecisionNode Edge](https://huggingface.co/bynn-intelligence-com/decisionnode-edge)** | Open weights | private and on-device deployment, from GPUs to laptops |
| **[DecisionNode Edge Mini](https://huggingface.co/bynn-intelligence-com/decisionnode-edge-mini)** | Open weights | CPUs and constrained devices |

**Which model to use.** For speed, start with DecisionNode Flash: it answers in milliseconds and keeps high accuracy, so it suits real-time decisions at any volume. When a decision needs the highest accuracy, use DecisionNode. To decide on your own hardware, use DecisionNode Edge, and send only its unsure questions to the hosted models.

## The DecisionNode platform

- **Typed and calibrated.** Six question types (choice, score, truth, number, point and box), each answered with calibrated probabilities that mean the same thing from one question to the next.
- **Boosters.** Specialist models that read each image before the decision: Human Age, AI-generated image detection, Face detection, Face recognition, OCR, Depth Vision and Segment Vision.
- **Knowledge.** Connect your own documents, and every decision draws on the passages it needs.
- **Fine-tuning.** Adapt DecisionNode to your own decisions and data.
- **Built for production.** Streaming sessions, batch jobs, uploads and inputs of up to 65,536 tokens.
- **Very low cost.** Pricing starts at a few cents per million input tokens, output tokens are free, and every account includes a free daily allowance.

[Create a free account](https://decisionnode.com/console/keys) to get an API key, or read the [documentation](https://decisionnode.com/docs).

## Open models

DecisionNode Edge and DecisionNode Edge Mini run entirely on your own hardware with the `decisionnode-edge` runtime (`pip install decisionnode-edge`), with optimized backends for NVIDIA GPUs, Apple silicon, x86 CPUs and llama.cpp devices. They are released under the AGPL-3.0, with a commercial license available from Bynn Intelligence for use in closed products and services.

The recommended setup on your own hardware, from GPUs to robots and drones, is a cascade: DecisionNode Edge decides on the device, and only questions below your confidence threshold go to DecisionNode Flash, then to DecisionNode, on the hosted API. If you prefer a fully managed service at very low cost, use the hosted API directly.

## Contact

Commercial licensing and partnerships: [licensing@bynn.com](mailto:licensing@bynn.com). Website: [decisionnode.com](https://decisionnode.com).