startlux-models/StartLux-Decision-v1.1-2B-Instant

Hugging Face
VISIONPricing:Input $0.32 / Cached $0.064 / Output $1.6Concurrent Unit Cost:1Model Size:2.3BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Oct 10, 2026License:cc-by-nc-4.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Warm

StartLux-Decision-v1.1-2B-Instant by StartLux Labs is a 2.3 billion parameter decision model with a 32768 token context length. It is designed to process a given state (text or JSON, with up to 8 images) and answer typed questions (choice, yes/no, or score) by returning probabilities for each option in a single forward pass, without generating text. This model excels at structured decision-making tasks, providing probabilistic outputs for classification-like queries.

Loading preview...

Overview

StartLux-Decision-v1.1-2B-Instant is a 2.3 billion parameter model developed by StartLux Labs, specifically engineered for structured decision-making. Unlike traditional generative LLMs, this model takes a state (text, JSON, or up to 8 images) and a set of typed questions (e.g., choice, noul for yes/no, score on a scale) and outputs probabilities for each option in a single forward pass. It does not generate free-form text, making it highly efficient for classification and decision-support applications.

Key Capabilities

  • Probabilistic Decision Outputs: Provides a probability for every option of every question, enabling nuanced decision analysis.
  • Multimodal Input: Accepts text, JSON, and up to 8 images as input for the 'state' description.
  • Structured Question Types: Supports choice (pick one from options), noul (yes/no), and score (level on a 2-10 scale) questions.
  • Efficient Inference: Designed for a single forward pass to return all probabilities, optimizing for speed and resource usage.
  • Bidirectional Prompt Attention: Utilizes a unique attention mechanism where prompt tokens attend to later prompt tokens, allowing the model to read the state with the question in view.

Good For

  • Automated Triage and Routing: Ideal for systems that need to classify inputs (e.g., customer support tickets) and route them based on predefined criteria and urgency.
  • Structured Data Analysis: When the goal is to extract specific, probabilistic answers from complex, structured or semi-structured data.
  • Applications Requiring Confidence Scores: Useful in scenarios where not just an answer, but also the model's confidence in that answer, is crucial for downstream processes.
  • Integration with System One API: Compatible with the System One /v1/systemone API format for requests and responses.

Finetunes
4 models