Tasks and inputs
d1-ultra is listed for Visual question answering, Structured Decisions. It accepts text, image and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
SyncDev model directory · Profile updated Nov 18, 2026
Deprecated
Deprecated on 16 November 2026 and superseded by d1.2-ultra. Liquid AI has not announced when the hosted endpoint stops answering.
d1-ultra is Liquid AI's largest decision model: API-only, it scores yes/no, choice and rating questions over text and images at $0.042 per 1M input tokens.
The specification score summarizes recorded features and limits. It is not a benchmark result.
Pricing conditions: $0.042 per 1M input tokens; images 1.5 tokens per 32x32 patch; each question billed as its own prompt; no output tokens. Indicative; confirm on your own account.
Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.
Accepts
Produces
A 1024x1024 photo costs 1,536 tokens with d1-ultra. At $0.042 per million input tokens, that is a fraction of a hundredth of a cent per image per question. Liquid AI's new decision model, released on 15 October 2026, is priced for checking things at volume: a circuit board against its spec, a damage claim against the customer's photo.
It is a separate, newer model from the original Liquid d1, which Liquid now calls its previous model.
Three question types, all returned as probabilities with no generated text:
Inputs can be text, images or both. Under the hood is an unnamed larger open-weight model with heavy Liquid-specific customisation, trained in particular to give better-calibrated probabilities than decision models that tend to overconfidence.
| Item | Detail |
|---|---|
| Endpoint | api.liquid.ai/decisions/v1/systemone, "model": "d1-ultra" |
| Gateways | Vercel AI Gateway (liquid/d1-ultra), OpenRouter |
| Price | $0.042 per 1M input tokens; no output tokens (indicative) |
| Images | 1.5 tokens per 32x32 patch |
| Context | 32,768 tokens (gateway listing) |
| Weights | Closed |
The trap is in how questions are billed. Each question is charged as its own prompt, including its text and every attached image. Ask five questions about one high-resolution photo and you pay for the photo five times. Group what you can into fewer, better-designed questions.
Liquid says d1-ultra beats d1 and Jev 1.14 across 112 evaluations and leads a public decision index by at least 8 points, with the biggest gains on knowledge tasks. A competitor's tables from 11 November put it at 68.14 on another decision index, behind SubQ Snap 1.2, and report that its top answer changed on 21.2% of questions when options were reordered. Option-order sensitivity is a real risk for any classifier. Shuffle your labels in testing.
Two loose ends. Liquid's post gives no context window; the 32,768 figure comes from the gateway listing, so confirm it with a long request. And Liquid originally called Vercel and OpenRouter availability planned; both now list it.
Pick an inspection task with photos, such as returns, claims or quality checks. Label 300 cases. Send each with one well-phrased question, record the probability, and plot precision against threshold. Then shuffle the option order and rerun to see how stable the answers are. Compare with Jev and Drex 2.0, or browse structured decisions.
Compare the model on the work you need to complete.
d1-ultra is listed for Visual question answering, Structured Decisions. It accepts text, image and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
Documented limits: 32.77K context. The listed route has starting rates of $0.042 input per million tokens. Measure a completed task using the prompts, documents and tools your application will use. Include reasoning and retries in the result.
Use the model identifier and account requirements for the selected API route. Confirm its request limits and output format with one complete task before connecting a larger workflow.
These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.
Microsoft
Microsoft-Decision-1.1 is a decision model for routing, triage and classification, returning calibrated probabilities. GA in Foundry from November 2026.
Liquid AI
Liquid AI d1.2 is a hosted omnimodal decision model for text, image and audio questions with calibrated answers, at $0.02 per million input tokens.
Liquid AI
Liquid AI d1.2-ultra is the top hosted decision model in Liquid's line, scoring 75 on the AA Decision Index, with audio support at $0.042 per 1M input.
Gemini Decision 1 powers Google's Decisions API, returning calibrated yes/no, choice and rubric answers instead of text, at $0.10 per 1M input tokens.
subquadratic
SubQ Snap 1.2 is Subquadratic's long-context decision model on linear sparse attention, tested to 256K tokens, priced at $0.04 per 1M input tokens.
Perplexity
pplx-decider-flash-v1.3 is Perplexity's small, half-price decision model, distilled from the 27B decider, for high-volume classification and span extraction.