Robots Atlas>ROBOTS ATLAS
D1 Ultra
AI Models

D1 Ultra

ultra
Liquid AI's multimodal decision model. Takes text and images plus typed questions and returns probabilities (yes/no, choice, score) without generating text.
✓ Active✓ Public accessMultimodalSpecialized AI
Context window
33K
tokens
Release date
15 October 2026
🏢Liquid AIProducer
Access:APIHostedDeployment:☁ Cloud

Overview

d1-ultra is a multimodal decision model developed by Liquid AI, released on 15 October 2026. Unlike generative models it does not produce text — given unstructured input (text, images, or both) and one or more questions, it returns answer probabilities in a single forward pass.

The model answers three question types: yes/no (a probability from 0 to 1), one label chosen from many, and a score on a scale. It targets hard, messy real-world tasks such as spotting defects or checking claims against photos, with emphasis on properly calibrated probabilities.

d1-ultra is based on a larger open-weight model with Liquid AI's own customizations. It supports 44 languages and, according to the company, ranked first on the Artificial Analysis Decision Index with a score of 65, leading all eight categories of its 112-evaluation suite. It is available via the Liquid API (the /v1/systemone endpoint) at $0.042 per million input tokens; the model generates no output tokens.

Classification
MultimodalSpecialized AI
Access & deployment
APIHosted
Cloud
Weights: Closed
Key parameters
📏 Context: 33K
📥 Input: text, image

Technical specification

Context window
33K
tokens
Modalities
⬇ Input
textimage
⬆ Output
structured_data

Capabilities and applications

Native model capabilities
Classification
Assigning an observation to one of predefined classes (binary or multi-class). Output: class label and optionally probabilities.
Category: other
Structured output
Producing data in structured formats such as JSON.
Category: structured_generation
Multimodal understanding
Category: multimodal
Image understanding
Analysing and interpreting the content of images.
Category: vision
Multilingual
Competence in many natural languages (from a few to over a hundred): understanding, generation, translation, and code-switching within a single conversation. Frontier models support a wide range of languages with comparable quality.
Category: language

Benchmark results

1 benchmark
Artificial Analysis Decision Index
65
📄 Oficjalny blog Liquid AI
Per Liquid AI, the model ranked #1 on the index and leads all 8 categories of its 112-evaluation suite.

Pricing

Sources and related pages