{"slug": "liquid-ai-d1-ultra", "title": "Liquid AI D1-Ultra", "summary": "Liquid AI released d1-ultra, a multimodal decision model it says outperforms all other decision models on the Artificial Analysis Decision Index by at least 8 points and beats Jev 1.14 across all tasks, based on 112 unique evaluations. The model answers noul (yes/no with a 0-1 probability), Choice (one label from many with per-label probabilities), and Score (a position on a scale) questions, reads text, images, or both, and is available on the Liquid API at $0.042 per million input tokens, billed on input tokens only with images counted at 1.5 tokens per 32×32-pixel patch. Liquid AI said d1-ultra will soon be available through Vercel and OpenRouter.", "body_md": "[News](https://www.liquid.ai/news)\n\n[Models](https://www.liquid.ai/news/models)\n\n# d1-ultra: Introducing the World’s Best Decision Model\n\nToday, we release d1-ultra, our new multimodal **decision model with state-of-the-art performance**. Most decision models work well on easy and common tasks. We built d1-ultra for harder, messier real-world use cases, like spotting a defect on a circuit board against its spec or checking a damage claim against the customer’s photo.\n\nd1-ultra is available today on the [Liquid API](https://console.liquid.ai/) at $0.042 per million input tokens.\n\n[Try on Liquid Console](https://console.liquid.ai/)\n\n[Try on d1 Playground](https://d1.liquid.ai/)\n\n## What d1-ultra does\n\nd1-ultra answers the same three question types as d1:\n\n- **noul** : a yes/no question, answered with a probability from 0 to 1.\n- **Choice** : one label from many, with a probability per label.\n- **Score** : a position on a scale.\n\n## Results\n\nCompared to d1, we based d1-ultra on a larger open-weight model with a series of Liquid-specific heavy customizations. It provides significant gains when combined with our training pipeline, especially on knowledge-related tasks.\n\nWe compared it to our previous model d1 and Jev 1.14 across 112 unique evaluations.\n\nAccuracy is not enough: decision models also output probabilities that must be properly calibrated. We realized that most decision models are overconfident in their responses. That’s why we specifically trained d1-ultra to produce more reliable probabilities.\n\nd1-ultra outperforms Jev 1.14 across all tasks, delivering better-calibrated answers and higher accuracy at the same price. To see how it works in practice, you can find more demos running on d1-ultra in the [d1 Playground](http://d1.liquid.ai/).\n\nd1-ultra also outperforms all other decision models [Artificial Analysis Decision Index](https://artificialanalysis.ai/evaluations/artificial-analysis-decision-index) by at least 8 points.\"\n\n## Availability and Pricing\n\nd1-ultra is available on the [Liquid AI API](http://console.liquid.ai/). Create your API key and start building today. d1-ultra will soon be available through [Vercel](https://vercel.com/ai-gateway/models/d1) and [OpenRouter](https://openrouter.ai/liquid/d1).\n\nThe model reads text, images, or both at the same time, like in this example:\n\n``` python\nimport base64, os, requests\n\nimage = base64.b64encode(open(\"board.jpg\", \"rb\").read()).decode()\n\nresponse = requests.post(\n    \"https://api.liquid.ai/decisions/v1/systemone\",\n    headers={\"Authorization\": f\"Bearer {os.environ['LIQUID_API_KEY']}\"},\n    json={\n        \"model\": \"d1-ultra\",\n        \"images\": [f\"data:image/jpeg;base64,{image}\"],\n        \"state\": \"Camera image of a circuit board on the production line.\",\n        \"questions\": \n            {\"defect\": {\n                \"type\": \"noul\", \n                \"instructions\": \"Does this circuit board have a defect?\"\n             }\n         },\n    },\n)\n\nprint(response.json()[\"answers\"][\"defect\"][\"noul\"])\n```\n\nThe full API reference is in the [decision models documentation](https://docs.liquid.ai/lfm/models/decision-models).\n\nd1-ultra is billed on input tokens only, with no output tokens. Images are counted as input tokens at the same rate as text: 1.5 tokens per 32×32-pixel patch, so a 1024×1024 image costs 1,536 tokens. Each question is billed as its own prompt, including its text and all images.\n\nd1-ultra is our most capable decision model so far. We’re committed to continuing our work on decision models across a range of sizes, with new features, higher decision quality, and lower latency, and we plan to provide regular updates.\n\n[Try on Liquid Console](https://console.liquid.ai/)\n\n[Read our docs](https://docs.liquid.ai/lfm/models/decision-models)\n\n## **Citation**\n\nFor citations, please use the following reference or BibTeX:", "url": "https://wpnews.pro/news/liquid-ai-d1-ultra", "canonical_source": "https://www.liquid.ai/blog/d1-ultra", "published_at": "2026-10-15 17:15:43+00:00", "updated_at": "2026-10-15 17:26:45.476083+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "ai-products", "ai-tools"], "entities": ["Liquid AI", "d1-ultra", "d1", "Jev 1.14", "Artificial Analysis Decision Index", "Liquid API", "Vercel", "OpenRouter"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/liquid-ai-d1-ultra", "markdown": "https://wpnews.pro/news/liquid-ai-d1-ultra.md", "text": "https://wpnews.pro/news/liquid-ai-d1-ultra.txt", "jsonld": "https://wpnews.pro/news/liquid-ai-d1-ultra.jsonld"}}