AI BriefingLiquid AI releases d1-ultra, a multimodal decision model with strong Pareto front performance
Key point
The closed-source model is available via API at $0.042 per million input tokens and outperforms other decision models by at least 8 points on the Artificial Analysis Decision Index.
Details
Liquid AI has released d1-ultra, a multimodal decision model that processes text, images, or both in a single forward pass to return calibrated probabilities for yes/no, multi-label, and scale questions without generating tokens. Unlike its predecessor d1, d1-ultra is built on a larger open-weight base with Liquid-specific customizations, targeting complex real-world tasks such as validating circuit boards against specs or assessing damage claims from photos.
Performance and Calibration
The company reports that d1-ultra outperforms both d1 and Jev 1.14 across 112 evaluations, with notable improvements in knowledge-related tasks. The model is described as better calibrated, addressing the tendency of decision models to be overconfident. On the Artificial Analysis Decision Index, d1-ultra exceeds other decision models by at least 8 points.
Availability and Pricing
- Status: Closed-source (though smaller d1 models remain open-source).
- Access: Available now via the Liquid AI API.
- Cost: $0.042 per million input tokens.
- Future Listings: Vercel and OpenRouter integrations are planned.
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.
Related
Xiaomi Releases Open-Weight Multimodal Model MiMo-V2.6-Pro… Higher Intelligence and Competitive Pricing Compared to Similar-Sized Models
Artificial Analysis Releases Small LLM Inference Benchmark for Mobile DevicesMicrosoft introduces Microsoft-Decision-1, a low-latency model for structured decision-making