# Model Council in Perplexity.
Send one question to multiple models, then combine consensus, disagreements, and source details. Focus on parallelism and comparison, not a sequential relay among models.
```motion-demo
{
"id": "sequence",
"title": "Overall choreography",
"src": "demos/sequence.html",
"duration": 41.72,
"status": "authored",
"description": "Motion and composition example",
"initialTime": 8,
"params": [
{
"key": "title",
"label": "Sample main copy",
"type": "text",
"value": "Model Council",
"maxLength": 60
},
{
"key": "accent",
"label": "Accent color",
"type": "color",
"value": "#2c8587"
}
],
"presets": [
{
"label": "Default content",
"values": {
"title": "Model Council",
"accent": "#2c8587"
}
},
{
"label": "OpenDesign adaptation",
"values": {
"title": "OpenDesign",
"accent": "#7050da"
}
}
]
}
```
## Split one question into three answers
Choose mode and models first, then show three processing states simultaneously after submitting one question. Advance each independently while keeping names and positions fixed; enter comparison only after answers converge.
## Read consensus and disagreement separately
Present stable conclusions in the consensus table before model positions in the disagreement table. Reuse column widths and model order so sources stay easy to find across transitions.
| Information layer | Visual arrangement |
| --- | --- |
| Parallel answers | Three parallel status rows with fixed model names |
| Consensus | Highlight shared conclusions and reduce duplication |
| Disagreement | Compare columns by the same criteria |
| Details | Retain the source model and expand the full answer |
## Keep the path visible in details
When moving from a table to one model's answer, preserve its marker and selection. Increase detail density as the table becomes secondary; restore the original column order when returning.

## Differentiate levels with restrained color
Use pale backgrounds, dark text, and teal selection. Serif may serve proposition headings, with sans serif for tables and model labels; avoid competing large brand-color areas, reserving color for current actions and comparisons.
## Change models or adapt to proposal reviews
When replacing model names, questions, and answers, update comparison criteria too. Three proposals or teams also fit, provided shared input→parallel results→comparison→source details remains intact.
## Color and typography
Serif proposition headings with sans serif interface text; generous whitespace and fine borders.
#fafaf7 · #171917 · #2c8587 · #dbe4df · #ffffff
## Elements and production rules
Mode menu / model list / parallel status row / consensus table / disagreement table / answer details
- Branch the same question to multiple models, then merge completed answers into a comparison table; do not reduce this to three serial steps.
- Consensus and disagreement are separate information layers; keep source answers accessible at the end.
## Content replacement
- Adapt to multi-option comparisons, preserving parallel generation→comparison→source-detail tracing.
## Production parameters
[Visuals and objects](style.json) · [Section timing](timeline.json) · [Position data](tracks.json) · [File manifest](manifest.json)
This specification is based on the reference video. The HTML provides editable choreography examples; each file identifies the source, precision and valid interval of its data.
```motion-board
{
"version": 1,
"brand": "Perplexity",
"id": "V082",
"palette": [
"#fafaf7",
"#171917",
"#2c8587",
"#dbe4df",
"#ffffff"
],
"typography": "Serif message headings with sans-serif interface text, ample negative space and thin borders",
"sample": "Model Council",
"objects": [
"Mode menu",
"Model list",
"Parallel status row",
"Consensus table",
"Disagreement table",
"Answer details"
],
"frames": [
{
"path": "assets/reference-1.jpg",
"timeSeconds": 0.0,
"frameId": "K0001",
"role": "reference-evidence",
"status": "measured"
},
{
"path": "assets/reference-2.jpg",
"timeSeconds": 8.0,
"frameId": "K0023",
"role": "reference-evidence",
"status": "measured"
},
{
"path": "assets/reference-3.jpg",
"timeSeconds": 11.0,
"frameId": "K0026",
"role": "reference-evidence",
"status": "measured"
},
{
"path": "assets/reference-4.jpg",
"timeSeconds": 21.0,
"frameId": "K0040",
"role": "reference-evidence",
"status": "measured"
},
{
"path": "assets/reference-5.jpg",
"timeSeconds": 25.0,
"frameId": "K0048",
"role": "reference-evidence",
"status": "measured"
},
{
"path": "assets/reference-6.jpg",
"timeSeconds": 36.0,
"frameId": "K0066",
"role": "reference-evidence",
"status": "measured"
}
],
"stages": [
{
"start": 0,
"label": "Prism heading",
"end": 4,
"status": "inferred",
"basis": "Representative-frame review; approximate boundaries"
},
{
"start": 4,
"label": "Choose mode",
"end": 8,
"status": "inferred",
"basis": "Representative-frame review; approximate boundaries"
},
{
"start": 8,
"label": "Choose three models",
"end": 11,
"status": "inferred",
"basis": "Representative-frame review; approximate boundaries"
},
{
"start": 11,
"label": "Ask",
"end": 15,
"status": "inferred",
"basis": "Representative-frame review; approximate boundaries"
},
{
"start": 15,
"label": "Parallel answers",
"end": 21,
"status": "inferred",
"basis": "Representative-frame review; approximate boundaries"
},
{
"start": 21,
"label": "Consensus table",
"end": 25,
"status": "inferred",
"basis": "Representative-frame review; approximate boundaries"
},
{
"start": 25,
"label": "Disagreement table",
"end": 29,
"status": "inferred",
"basis": "Representative-frame review; approximate boundaries"
},
{
"start": 29,
"label": "Single-model details",
"end": 36,
"status": "inferred",
"basis": "Representative-frame review; approximate boundaries"
},
{
"start": 36,
"label": "Closing",
"end": 41.72,
"status": "inferred",
"basis": "Representative-frame review; approximate boundaries"
}
],
"duration": 41.72,
"fontFamily": "Georgia, serif"
}
```