ui : put the draft sidecar on the model id and its details in the pane

Assisted-by: pi:llama.cpp/DeepSeek-V4.1-Flash

tools/ui/src/lib/components/app/models/ModelDraftChip.svelte

@@ -1,41 +0,0 @@
-<script lang="ts">
-	import ModelAvatar from './ModelAvatar.svelte';
-	import type { ModelDraft } from './ModelsManager/utils';
-	import { Zap } from '@lucide/svelte';
-	import { Badge } from '$lib/components/ui/badge';
-	import type { ModelOption } from '$lib/types/models';
-	import { rawModelId } from '$lib/utils/model-option-id';
-
-	interface Props {
-		/** The draft model's option, when the manager lists it, so its avatar can be shown. */
-		option?: ModelOption | null;
-		draft: ModelDraft;
-	}
-
-	let { draft, option = null }: Props = $props();
-</script>
-
-{#if draft.active}
-	<Badge
-		class="h-5 max-w-40 gap-1 px-1.5 text-[10px]"
-		title="Speculative draft in use"
-		variant="secondary"
-	>
-		<Zap class="h-3 w-3 shrink-0" />
-
-		{#if option}
-			<ModelAvatar {option} showRepoOrgAvatar size="size-3.5" />
-		{/if}
-
-		<span class="truncate">{rawModelId(option?.model ?? draft.model ?? draft.kind ?? 'draft')}</span
-		>
-	</Badge>
-{:else}
-	<Badge
-		class="h-5 gap-1 px-1.5 text-[10px] text-muted-foreground/70"
-		title="Draft sidecar in this repo, not in use"
-		variant="outline"
-	>
-		{draft.kind}
-	</Badge>
-{/if}

tools/ui/src/lib/components/app/models/ModelId.svelte

@@ -1,9 +1,11 @@
 <script lang="ts">
 	import ModelCapabilityIcons from './ModelCapabilityIcons.svelte';
+	import type { ModelDraft } from './ModelsManager/utils';
 	import { Database, ScrollText } from '@lucide/svelte';
 	import { TruncatedText } from '$lib/components/app';
 	import * as Tooltip from '$lib/components/ui/tooltip';
 	import { type ModelSidecar } from '$lib/constants';
+	import { MODEL_BADGE_CLASS, MODEL_VARIANT_BADGE_CLASS } from '$lib/constants';
 	import { HuggingFaceService } from '$lib/services';
 	import { ModelsService } from '$lib/services/models.service';
 	import { settingsStore } from '$lib/stores';
@@ -36,6 +38,8 @@
 		/** Min/max GGUF file size (main + draft) across quants; renders a range when set. */
 		sizeRange?: { min: number; max: number } | null;
 		draftSidecars?: ModelSidecar[];
+		/** Drafts a load would use, rendered as `+ [KIND] [QUANT]` next to the id. */
+		drafts?: ModelDraft[];
 		/** Allow badges to wrap onto new lines instead of truncating. */
 		wrap?: boolean;
 		class?: string;
@@ -45,6 +49,7 @@
 		aliases,
 		class: className = '',
 		contextLength,
+		drafts = [],
 		draftSidecars = [],
 		hideCapabilities = false,
 		hideModalities = false,
@@ -67,12 +72,10 @@
 		...rest
 	}: Props = $props();
 
-	const badgeClass =
-		'inline-flex w-fit shrink-0 items-center justify-center whitespace-nowrap rounded-md border border-border/50 px-1 py-0 text-[10px] font-mono bg-foreground/15 dark:bg-foreground/10 text-foreground [a&]:hover:bg-foreground/25';
+	const badgeClass = MODEL_BADGE_CLASS;
 	const tagBadgeClass =
 		'inline-flex w-fit shrink-0 items-center justify-center whitespace-nowrap rounded-md border border-border/50 px-1 py-0 text-[10px] font-mono text-foreground [a&]:hover:bg-accent [a&]:hover:text-accent-foreground';
-	const variantBadgeClass =
-		'inline-flex w-fit shrink-0 items-center justify-center whitespace-nowrap rounded-md bg-primary px-1.5 py-0 text-[10px] font-mono font-semibold uppercase tracking-wide text-primary-foreground';
+	const variantBadgeClass = MODEL_VARIANT_BADGE_CLASS;
 
 	/** Alias badges beyond this many collapse into a single `+x more` badge. */
 	const MAX_ALIAS_BADGES = 2;
@@ -89,6 +92,7 @@
 	let uniqueAliases = $derived([...new Set(aliases ?? [])]);
 	let uniqueTags = $derived([...new Set([...(parsed.tags ?? []), ...(tags ?? [])])]);
 	let uniqueDraftSidecars = $derived([...new Set(draftSidecars)].filter((s) => !isAuxSidecar(s)));
+	let activeDrafts = $derived(drafts.filter((draft) => draft.active && draft.kind));
 
 	let primaryAlias = $derived(uniqueAliases.length === 1 ? uniqueAliases[0] : null);
 	let displayName = $derived(primaryAlias ?? parsed.modelName ?? modelId);
@@ -131,6 +135,18 @@
 					</span>
 				{/if}
 
+				{#each activeDrafts as draft (draft.kind ?? 'draft')}
+					<span class="flex shrink-0 items-center gap-1">
+						<span class="text-[10px] text-muted-foreground">+</span>
+
+						<span class={variantBadgeClass} title="Speculative draft in use">{draft.kind}</span>
+
+						{#if draft.quant}
+							<span class={badgeClass}>{draft.quant}</span>
+						{/if}
+					</span>
+				{/each}
+
 				{#each uniqueDraftSidecars as sidecar (sidecar)}
 					<span class={variantBadgeClass} title={`${sidecar.toUpperCase()} draft model available`}>
 						{sidecar}

tools/ui/src/lib/components/app/models/ModelsManager/ModelsManagerModelConfiguration/ModelsManagerModelConfiguration.svelte

@@ -108,7 +108,12 @@
 
 		<div class="min-h-0 flex-1 overflow-y-auto px-4 py-4">
 			<Tabs.Content value="information">
-				<ModelsManagerModelConfigurationInformation hub={hubDetails} {option} {serverProps} />
+				<ModelsManagerModelConfigurationInformation
+					draftSetting={override?.load?.speculativeDecoding ?? null}
+					hub={hubDetails}
+					{option}
+					{serverProps}
+				/>
 			</Tabs.Content>
 
 			<Tabs.Content value="load">

tools/ui/src/lib/components/app/models/ModelsManager/ModelsManagerModelConfiguration/ModelsManagerModelConfigurationInformation.svelte

@@ -1,5 +1,11 @@
 <script lang="ts">
-	import { isLocalOption, modelQuantLabel, modelSizeLabel, resolveModelSize } from '../utils';
+	import {
+		isLocalOption,
+		modelDraftsFor,
+		modelQuantLabel,
+		modelSizeLabel,
+		resolveModelSize
+	} from '../utils';
 	import {
 		ActionIconCopyToClipboard,
 		BadgesModality,
@@ -14,13 +20,15 @@
 	import { formatFileSize, formatNumber, formatParameters } from '$lib/utils/formatters';
 
 	interface Props {
+		/** Draft the load settings name, when one is set. */
+		draftSetting?: string | null;
 		/** Hugging Face details, filled in when discovery is on and the server has none. */
 		hub?: HfModelDetailInfo | null;
 		option: ModelOption;
 		serverProps: ApiLlamaCppServerProps | null | undefined;
 	}
 
-	let { hub = null, option, serverProps }: Props = $props();
+	let { draftSetting = null, hub = null, option, serverProps }: Props = $props();
 
 	let quant = $derived(modelQuantLabel(option));
 	let size = $derived(modelSizeLabel(option));
@@ -55,6 +63,10 @@
 				? `${formatNumber(hub.gguf.context_length)} tokens`
 				: null
 	);
+	// the draft a load would speculate with, and any other sidecar on disk
+	let drafts = $derived(modelDraftsFor(option, draftSetting));
+	let activeDraft = $derived(drafts.find((draft) => draft.active) ?? null);
+	let idleDrafts = $derived(drafts.filter((draft) => !draft.active));
 	let meta = $derived(option.meta);
 	// the server reports these once the model is loaded; the Hub knows them anyway
 	let gguf = $derived(hub?.gguf ?? null);
@@ -93,6 +105,22 @@
 			value: (option.meta?.architecture as string) ?? gguf?.architecture ?? null
 		},
 		{ label: 'Format', value: isLocalOption(option) ? 'GGUF' : null },
+		...(activeDraft
+			? [
+					{ isBadge: true, label: 'Draft sidecar', value: activeDraft.kind },
+					{
+						isCopyable: true,
+						label: 'Draft model',
+						value: activeDraft.model ?? option.model.split(':')[0]
+					},
+					{ isBadge: true, label: 'Draft quant', value: activeDraft.quant }
+				]
+			: []),
+		{
+			isBadge: true,
+			label: 'Other sidecars',
+			value: idleDrafts.length > 0 ? idleDrafts.map((draft) => draft.kind).join(', ') : null
+		},
 		{ label: 'Parallel Slots', value: serverProps ? String(serverProps.total_slots) : null },
 		{ isMono: true, label: 'Build Info', value: serverProps?.build_info ?? null }
 	] satisfies Array<{

tools/ui/src/lib/components/app/models/ModelsManager/ModelsManagerModelsTable.svelte

@@ -1,8 +1,8 @@
 <script lang="ts">
-	import ModelDraftChip from '../ModelDraftChip.svelte';
 	import ModelsManagerFilters from './ModelsManagerFilters.svelte';
 	import {
 		draftFromArgs,
+		draftFromSetting,
 		isLocalOption,
 		type ModalityKey,
 		modelContextLength,
@@ -10,7 +10,8 @@
 		modelDrafts,
 		type ModelOverride,
 		type ModelQuantGroup,
-		type ModelsTableGroup
+		type ModelsTableGroup,
+		sidecarFilesFor
 	} from './utils';
 	import {
 		ArrowDown,
@@ -46,20 +47,13 @@
 	import ModelsSelectorDownloadItem from '$lib/components/app/models/ModelsSelector/ModelsSelectorDownloadItem.svelte';
 	import { Badge } from '$lib/components/ui/badge';
 	import { Button } from '$lib/components/ui/button';
-	import { FAMILY_ROW_WINDOW, LOCAL_BACKEND_ID, MODEL_ROW_WINDOW } from '$lib/constants';
+	import { FAMILY_ROW_WINDOW, MODEL_ROW_WINDOW } from '$lib/constants';
 	import { ModelDownloadConfirmAction, ServerModelStatus } from '$lib/enums';
-	import {
-		backendsModelsStore,
-		backendsStore,
-		modelsStore,
-		settingsStore,
-		uiStore
-	} from '$lib/stores';
+	import { backendsStore, modelsStore, settingsStore, uiStore } from '$lib/stores';
 	import type { ModelOption } from '$lib/types/models';
 	import { getBackend } from '$lib/utils/api-base';
 	import { getBackendCapabilities } from '$lib/utils/backend';
 	import { groupModelFamilies, type ModelFamilyGroup } from '$lib/utils/model-families';
-	import { rawModelId } from '$lib/utils/model-option-id';
 	import { SvelteSet } from 'svelte/reactivity';
 
 	interface Props {
@@ -227,32 +221,13 @@
 		// speculative decoding is a llama.cpp feature
 		if (!getBackendCapabilities(getBackend(option.backendId)).loadUnload) return [];
 
-		const repo = option.model.split(':')[0] ?? '';
-		const state = backendsModelsStore.get(option.backendId ?? LOCAL_BACKEND_ID);
 		// the router reports the arguments a model loads with, draft included
-		const serverDraft = draftFromArgs(
-			modelsStore.routerModels.find((model) => model.id === option.model)?.status?.args
-		);
+		const args = modelsStore.routerModels.find((model) => model.id === option.model)?.status?.args;
 		const configured =
-			serverDraft?.model ?? overrides[option.id]?.load?.speculativeDecoding ?? null;
+			draftFromArgs(args, option) ??
+			draftFromSetting(option, overrides[option.id]?.load?.speculativeDecoding);
 
-		return modelDrafts(option, state.drafts?.[repo] ?? [], configured, serverDraft?.kind ?? null);
-	}
-
-	/** The draft model's own option, when the manager lists it, for its avatar. */
-	function draftOptionFor(draft: ModelDraft): ModelOption | null {
-		const id = draft.model;
-
-		if (!id) return null;
-
-		const raw = rawModelId(id);
-
-		// a draft may be named as a repo, in which case any of its quants will do
-		return (
-			modelsStore.models.find((option) => option.id === id) ??
-			modelsStore.models.find((option) => rawModelId(option.id).startsWith(`${raw}:`)) ??
-			null
-		);
+		return modelDrafts(option, sidecarFilesFor(option), configured);
 	}
 
 	/** Name of the model the configuration pane has open, for the draft action label. */
@@ -405,6 +380,7 @@
 				<ModelId
 					aliases={option.aliases}
 					class="min-w-0 flex-1"
+					drafts={draftsFor(option)}
 					hideCapabilities
 					modalities={option.modalities}
 					modelId={option.model}
@@ -413,10 +389,6 @@
 				/>
 
 				<ModelCapabilities {option} />
-
-				{#each draftsFor(option) as draft (draft.model ?? draft.kind)}
-					<ModelDraftChip {draft} option={draftOptionFor(draft)} />
-				{/each}
 			</span>
 		</span>
 
@@ -471,6 +443,7 @@
 					<ModelId
 						aliases={entry.base.aliases}
 						class="min-w-0"
+						drafts={draftsFor(entry.base)}
 						hideCapabilities
 						hideQuantization
 						modalities={mediaSource.modalities}
@@ -480,10 +453,6 @@
 					/>
 
 					<ModelCapabilities option={entry.base} />
-
-					{#each draftsFor(entry.base) as draft (draft.model ?? draft.kind)}
-						<ModelDraftChip {draft} option={draftOptionFor(draft)} />
-					{/each}
 				</span>
 
 				<span class="block text-xs text-muted-foreground">{groupLabel}</span>

tools/ui/src/lib/components/app/models/ModelsManager/utils.ts

@@ -7,8 +7,13 @@ import {
 	SPEC_TYPE
 } from '$lib/constants';
 import { HuggingFaceService, ModelsService } from '$lib/services';
-import { settingsStore } from '$lib/stores';
-import type { ModelLoadProgress, ModelModalities, ModelOption } from '$lib/types/models';
+import { backendsModelsStore, modelsStore, settingsStore } from '$lib/stores';
+import type {
+	ModelLoadProgress,
+	ModelModalities,
+	ModelOption,
+	ModelSidecarFile
+} from '$lib/types/models';
 import { getBackend } from '$lib/utils/api-base';
 import { formatFileSize, formatParameters } from '$lib/utils/formatters';
 import { rawModelId } from '$lib/utils/model-option-id';
@@ -121,13 +126,15 @@ export function saveOverrides(overrides: ModelOverrideMap): void {
 }
 
 /** Backend a model is served by, the local server reads as "This server". */
-/** A draft a model can speculate with, and whether the load settings point at it. */
+/** A draft a model can speculate with, and whether a load would use it. */
 export interface ModelDraft {
-	/** The draft the settings name, which is the one a load would use. */
+	/** The draft a load would use, from the server's own arguments or from the settings. */
 	active: boolean;
 	kind: ModelSidecar | null;
-	/** Qualified id of a draft from another repo; null when the file sits in this repo. */
+	/** Repo the draft comes from; null when the file sits in the model's own repo. */
 	model: string | null;
+	params: string | null;
+	quant: string | null;
 }
 
 /** Repo an id belongs to: the id without its quant tag. */
@@ -166,19 +173,64 @@ function repoFromDraftPath(path: string): string {
  * Draft the server's own launch arguments point at. The router reports the arguments a
  * model loads with, so this is what a load would really speculate with.
  */
-export function draftFromArgs(args: string[] | undefined): ModelDraft | null {
+export function draftFromArgs(args: string[] | undefined, option: ModelOption): ModelDraft | null {
 	const flag = args?.indexOf('--model-draft') ?? -1;
 	const path = flag === -1 ? null : (args?.[flag + 1] ?? null);
 
 	if (!path) return null;
 
+	const parsed = ModelsService.parseModelId(path.split(/[/\\]/).pop() ?? path);
+	const repo = repoFromDraftPath(path);
+
 	return {
 		active: true,
-		kind: sidecarFromSpecType(args?.[(args?.indexOf('--spec-type') ?? -1) + 1]),
-		model: repoFromDraftPath(path)
+		kind: sidecarFromSpecType(args?.[(args?.indexOf('--spec-type') ?? -1) + 1]) ?? parsed.sidecar,
+		model: repo === repoOf(option.model) ? null : repo,
+		params: parsed.params
+			? `${parsed.params}${parsed.activatedParams ? `-${parsed.activatedParams}` : ''}`
+			: null,
+		quant: parsed.quantization
 	};
 }
 
+/** Draft the load settings name, resolved against the model's own repo. */
+export function draftFromSetting(option: ModelOption, value?: string | null): ModelDraft | null {
+	const id = value?.trim();
+
+	if (!id || id === 'off') return null;
+
+	const parsed = ModelsService.parseModelId(id);
+
+	return {
+		active: true,
+		kind: parsed.sidecar,
+		model: repoOf(id) === repoOf(option.model) ? null : id,
+		params: parsed.params
+			? `${parsed.params}${parsed.activatedParams ? `-${parsed.activatedParams}` : ''}`
+			: null,
+		quant: parsed.quantization
+	};
+}
+
+/**
+ * Drafts of a model, in the order they matter: what the server loads with, else what the
+ * settings name, then any other sidecar the model's own repo ships.
+ */
+export function modelDraftsFor(option: ModelOption, settingValue?: string | null): ModelDraft[] {
+	const args = modelsStore.routerModels.find((model) => model.id === option.model)?.status?.args;
+	const configured = draftFromArgs(args, option) ?? draftFromSetting(option, settingValue);
+
+	return modelDrafts(option, sidecarFilesFor(option), configured);
+}
+
+/** Draft sidecars a listing reported for the model's repo. */
+export function sidecarFilesFor(option: ModelOption): ModelSidecarFile[] {
+	const repo = option.model.split(':')[0] ?? '';
+	const state = backendsModelsStore.get(option.backendId ?? LOCAL_BACKEND_ID);
+
+	return state.drafts?.[repo] ?? [];
+}
+
 /**
  * Drafts to show for a model: the one the load settings name, plus any draft sidecar
  * the model's own repo ships. The configured one is the active draft; a sidecar that
@@ -186,32 +238,26 @@ export function draftFromArgs(args: string[] | undefined): ModelDraft | null {
  */
 export function modelDrafts(
 	option: ModelOption,
-	available: ModelSidecar[] = [],
-	configured?: string | null,
-	configuredKind?: ModelSidecar | null
+	available: ModelSidecarFile[] = [],
+	configured?: ModelDraft | null
 ): ModelDraft[] {
-	const value = configured?.trim() ?? '';
 	const drafts: ModelDraft[] = [];
 
-	if (value && value !== 'off') {
-		const parsed = ModelsService.parseModelId(value);
-		const sameRepo = repoOf(value) === repoOf(option.model);
+	if (configured) drafts.push(configured);
+
+	for (const file of available) {
+		// a sidecar a load already points at is the active draft, not a second entry
+		if (drafts.some((draft) => draft.kind === file.kind)) continue;
 
 		drafts.push({
-			active: true,
-			kind: parsed.sidecar ?? configuredKind ?? null,
-			model: sameRepo ? null : value
+			active: false,
+			kind: file.kind,
+			model: null,
+			params: file.params,
+			quant: file.quant
 		});
 	}
 
-	for (const kind of available) {
-		const alreadyActive = drafts.some((draft) => draft.active && draft.kind === kind);
-
-		if (alreadyActive) continue;
-
-		drafts.push({ active: false, kind, model: null });
-	}
-
 	return drafts;
 }
 

tools/ui/src/lib/components/app/models/ModelsSelector/ModelsSelectorOption.svelte

@@ -1,5 +1,6 @@
 <script lang="ts">
 	import ModelLoadHighlight from '../ModelLoadHighlight.svelte';
+	import { modelDraftsFor } from '../ModelsManager/utils';
 	import { ModelAvatar, ModelId, ModelLoadControl, ModelRowActions } from '$lib/components/app';
 	import { ModelCapability, ServerModelStatus } from '$lib/enums';
 	import { modelsStore } from '$lib/stores';
@@ -94,6 +95,7 @@
 	<ModelId
 		aliases={option.aliases}
 		class="min-w-0 flex-1"
+		drafts={modelDraftsFor(option)}
 		{hideOrgName}
 		{modalities}
 		modelId={option.model}

tools/ui/src/lib/constants/css-classes.constants.ts

@@ -29,6 +29,14 @@ export const CHAT_INPUT_FOCUS_SELECTOR =
  */
 export const DROPDOWN_MENU_CONTENT_SEARCH_SELECTOR = '[data-slot="dropdown-menu-content"] input';
 
+/** Neutral model badge: params, quantization, tags. */
+export const MODEL_BADGE_CLASS =
+	'inline-flex w-fit shrink-0 items-center justify-center whitespace-nowrap rounded-md border border-border/50 px-1 py-0 text-[10px] font-mono bg-foreground/15 dark:bg-foreground/10 text-foreground [a&]:hover:bg-foreground/25';
+
+/** Emphasis model badge: draft sidecars and other markers that must stand out. */
+export const MODEL_VARIANT_BADGE_CLASS =
+	'inline-flex w-fit shrink-0 items-center justify-center whitespace-nowrap rounded-md bg-primary px-1.5 py-0 text-[10px] font-mono font-semibold uppercase tracking-wide text-primary-foreground';
+
 /** Default Tailwind size class for inline icon components (lucide, etc.). */
 export const ICON_CLASS_DEFAULT = 'h-4 w-4';
 

tools/ui/src/lib/services/models.service.ts

@@ -14,7 +14,7 @@ import {
 	SIDECAR_TOKENS
 } from '$lib/constants';
 import { ServerModelStatus } from '$lib/enums';
-import type { ParsedModelId } from '$lib/types/models';
+import type { ModelSidecarFile, ParsedModelId } from '$lib/types/models';
 import {
 	apiDelete,
 	apiFetch,
@@ -40,6 +40,13 @@ function sidecarTokenInFilename(modelId: string): ModelSidecar | null {
 	return (match as ModelSidecar | undefined) ?? null;
 }
 
+/** Parameter label a parsed id reports, e.g. `35B-A3B`. */
+function paramsLabel(parsed: ParsedModelId): string | null {
+	if (!parsed.params) return null;
+
+	return `${parsed.params}${parsed.activatedParams ? `-${parsed.activatedParams}` : ''}`;
+}
+
 export class ModelsService {
 	private static readonly SSE_RECONNECT_MS = 1000;
 
@@ -117,27 +124,35 @@ export class ModelsService {
 	 * belong to. The router lists a downloaded sidecar as a model of its own, so this
 	 * keeps the pairing that the model list itself is filtered to drop.
 	 */
-	static draftSidecarsByRepo(response: ApiModelsListResponse): Record<string, ModelSidecar[]> {
-		const byRepo: Record<string, ModelSidecar[]> = {};
+	static draftSidecarsByRepo(response: ApiModelsListResponse): Record<string, ModelSidecarFile[]> {
+		const byRepo: Record<string, ModelSidecarFile[]> = {};
 
 		for (const entry of response.data ?? []) {
-			const sidecar =
-				ModelsService.parseModelId(entry.id).sidecar ?? sidecarTokenInFilename(entry.id);
+			const parsed = ModelsService.parseModelId(entry.id);
+			const sidecar = parsed.sidecar ?? sidecarTokenInFilename(entry.id);
 
 			if (!sidecar || isAuxSidecar(sidecar)) continue;
 
 			// the repo is the id with its quant tag and sidecar token taken off; naming
 			// parts of the id are not touched, a model name may carry `-4b` for instance
-			const repo = entry.id
+			const model = entry.id
 				.split(MODEL_ID.QUANTIZATION_SEPARATOR)[0]
 				.replace(MODEL_ID.WEIGHT_EXTENSION_REGEX, '')
 				.replace(new RegExp(`[-_ ]?${sidecar}([-_ ]?draft)?$`, 'i'), '');
 
-			if (!repo) continue;
+			if (!model) continue;
 
-			const kinds = (byRepo[repo] ??= []);
+			const files = (byRepo[model] ??= []);
 
-			if (!kinds.includes(sidecar)) kinds.push(sidecar);
+			if (files.some((file) => file.kind === sidecar)) continue;
+
+			files.push({
+				id: entry.id,
+				kind: sidecar,
+				model,
+				params: paramsLabel(parsed),
+				quant: parsed.quantization
+			});
 		}
 
 		return byRepo;

tools/ui/src/lib/stores/backendsModels.svelte.ts

@@ -7,16 +7,16 @@
  * cache is the prefetch layer the switches start from.
  */
 
-import type { ModelSidecar } from '$lib/constants';
 import { BackendsService } from '$lib/services/backends.service';
 import { ModelsService } from '$lib/services/models.service';
 import { backendsStore } from '$lib/stores/backends.svelte';
 import type { ApiModelsListResponse } from '$lib/types';
+import type { ModelSidecarFile } from '$lib/types/models';
 import type { ModelOption } from '$lib/types/models';
 
 export interface BackendModelsState {
 	/** Draft sidecars the listing carries, keyed by the repo they belong to. */
-	drafts?: Record<string, ModelSidecar[]>;
+	drafts?: Record<string, ModelSidecarFile[]>;
 	error: string | null;
 	loaded: boolean;
 	loading: boolean;

tools/ui/src/lib/types/models.d.ts

@@ -51,6 +51,21 @@ export interface ModelDownloadProgress {
 	files: Record<string, ModelDownloadFileProgress>;
 }
 
+/**
+ * A draft sidecar file a model listing reports as its own entry. The router lists a
+ * downloaded sidecar as a model, so this is what pairs it back with its model.
+ */
+export interface ModelSidecarFile {
+	id: string;
+	kind: ModelSidecar;
+	/** Repo the sidecar belongs to, e.g. `ggml-org/Qwen3.6-35B-A3B-GGUF`. */
+	model: string;
+	/** Parameter count the sidecar reports, e.g. `35B-A3B`. */
+	params: string | null;
+	/** Quantization of the sidecar file, e.g. `Q4_0`. */
+	quant: string | null;
+}
+
 export interface ParsedModelId {
 	raw: string;
 	orgName: string | null;