AI BriefingLiquid AI's d1-ultra decision model now available on AI Gateway
Key point
The multimodal model returns structured probabilities for classification and routing tasks rather than generating text.
Details
Liquid AI's d1-ultra is now accessible via AI Gateway as a multimodal decision model designed for classification, routing, and scoring across text and images. Unlike generative models, d1-ultra returns structured answers with probabilities, supporting Boolean questions, named category selection, and ordered scale scoring.
Core Capabilities and Pricing
The model supports a 32,768-token context window and is billed exclusively on input tokens, including image inputs, with no charges for output tokens. It is optimized for real-world decisions requiring calibrated probabilities, such as visual defect inspection or verifying damage claims against customer photos. Users can query multiple questions about a single input in one request to determine whether to automate an action or escalate to human review.
Integration and Billing
Developers can access the model using the name liquid/d1-ultra through the AI SDK, OpenAI-compatible Decisions API, or TypeSafe-compatible API. AI Gateway provides a unified interface for tracking usage and costs, reflecting provider pricing with no markup or platform fees on inference, including Bring Your Own Key requests.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.