@oh-my-pi/pi-catalog 15.11.7 → 15.11.8

package/CHANGELOG.md

@@ -2,6 +2,12 @@
 
 ## [Unreleased]
 
+## [15.11.8] - 2026-06-12
+
+### Fixed
+
+- Fixed Antigravity `gemini-3.1-pro --thinking high` failing with `Cloud Code Assist API error (400): Request contains an invalid argument.` — the upstream `gemini-3.1-pro-high` deployment rejects every `streamGenerateContent` request on both CCA endpoints while discovery still advertises it. High effort now routes to `gemini-pro-agent` (the same "Gemini 3.1 Pro (High)" model, verified accepting the identical request body), and the model-cache fingerprint version was bumped (`merge-v2` → `merge-v3`) so existing fresh caches refetch discovery and pick up the corrected routing immediately.
+
 ## [15.11.7] - 2026-06-12
 ### Added
 

package/dist/types/variant-collapse.d.ts

@@ -21,6 +21,15 @@ export interface EffortVariantFamily {
      * when it equals the logical id).
      */
     members: readonly string[];
+    /**
+     * Wire ids upstream no longer serves (e.g. a deployment killed while
+     * discovery still advertises it). Fresh collapsing never routes to them,
+     * and stale collapsed snapshots (bundled catalog, cache rows,
+     * previous-generation fallbacks) get routing/`requestModelId` entries that
+     * target them re-pointed through `routing`. Keep retired ids in `members`
+     * so the raw upstream spec is still consumed and aliased.
+     */
+    retiredMembers?: readonly string[];
     /**
      * Per-effort upstream wire id; `"off"` applies when thinking is disabled.
      * Entries whose target member is absent from the input are dropped — those

package/package.json

@@ -1,7 +1,7 @@
 {
 	"type": "module",
 	"name": "@oh-my-pi/pi-catalog",
-	"version": "15.11.7",
+	"version": "15.11.8",
 	"description": "Model catalog for omp: bundled model database, provider discovery descriptors, model identity, classification, and equivalence",
 	"homepage": "https://omp.sh",
 	"author": "Can Boluk",
@@ -34,11 +34,11 @@
 	},
 	"dependencies": {
 		"@bufbuild/protobuf": "^2.12.0",
-		"@oh-my-pi/pi-utils": "15.11.7",
+		"@oh-my-pi/pi-utils": "15.11.8",
 		"zod": "4.4.3"
 	},
 	"devDependencies": {
-		"@oh-my-pi/pi-ai": "15.11.7",
+		"@oh-my-pi/pi-ai": "15.11.8",
 		"@types/bun": "^1.3.14"
 	},
 	"engines": {

package/src/model-manager.ts

@@ -301,7 +301,7 @@ function retainModelIds<TApi extends Api>(
  * arms calling `resolveProviderModels` with the same `staticModels` array)
  * skip the JSON+hash work after the first call.
  */
-const MODEL_CACHE_FINGERPRINT_VERSION = "merge-v2";
+const MODEL_CACHE_FINGERPRINT_VERSION = "merge-v3";
 const kStaticFingerprint = Symbol("model-manager.staticFingerprint");
 type ModelArrayWithFingerprint = readonly Model<Api>[] & { [kStaticFingerprint]?: string };
 function fingerprintStatic<TApi extends Api>(

package/src/models.json

@@ -16574,7 +16574,7 @@
 				"effortRouting": {
 					"off": "gemini-3.1-pro-low",
 					"low": "gemini-3.1-pro-low",
-					"high": "gemini-3.1-pro-high"
+					"high": "gemini-pro-agent"
 				},
 				"suppressWhenOff": true
 			},
@@ -54417,7 +54417,7 @@
 				"cacheRead": 0.135,
 				"cacheWrite": 0
 			},
-			"contextWindow": 163840,
+			"contextWindow": 131072,
 			"maxTokens": 16384,
 			"thinking": {
 				"mode": "effort",

package/src/variant-collapse.ts

@@ -57,6 +57,15 @@ export interface EffortVariantFamily {
 	 * when it equals the logical id).
 	 */
 	members: readonly string[];
+	/**
+	 * Wire ids upstream no longer serves (e.g. a deployment killed while
+	 * discovery still advertises it). Fresh collapsing never routes to them,
+	 * and stale collapsed snapshots (bundled catalog, cache rows,
+	 * previous-generation fallbacks) get routing/`requestModelId` entries that
+	 * target them re-pointed through `routing`. Keep retired ids in `members`
+	 * so the raw upstream spec is still consumed and aliased.
+	 */
+	retiredMembers?: readonly string[];
 	/**
 	 * Per-effort upstream wire id; `"off"` applies when thinking is disabled.
 	 * Entries whose target member is absent from the input are dropped — those
@@ -123,12 +132,18 @@ export const ANTIGRAVITY_VARIANT_COLLAPSE_TABLE: VariantCollapseTable = {
 		{
 			id: "gemini-3.1-pro",
 			name: "Gemini 3.1 Pro",
-			// `gemini-pro-agent` is consumed as a member but unused in routing.
-			members: ["gemini-3.1-pro-low", "gemini-3.1-pro-high", "gemini-pro-agent"],
+			// High routes to `gemini-pro-agent` — the upstream `gemini-3.1-pro-high`
+			// deployment returns INVALID_ARGUMENT on every streamGenerateContent
+			// request (both CCA endpoints) while discovery still lists it;
+			// `gemini-pro-agent` is the same model ("Gemini 3.1 Pro (High)", same
+			// thinking budget/caps) and accepts the identical request body.
+			// `gemini-3.1-pro-high` stays a member so the dead raw id is consumed.
+			members: ["gemini-3.1-pro-low", "gemini-pro-agent", "gemini-3.1-pro-high"],
+			retiredMembers: ["gemini-3.1-pro-high"],
 			routing: {
 				off: "gemini-3.1-pro-low",
 				[Effort.Low]: "gemini-3.1-pro-low",
-				[Effort.High]: "gemini-3.1-pro-high",
+				[Effort.High]: "gemini-pro-agent",
 			},
 			thinking: { mode: "google-level", efforts: GEMINI_3_PRO_FAMILY_EFFORTS },
 			suppressWhenOff: true,
@@ -282,6 +297,69 @@ export function isVariantCollapsedSpec(spec: VariantSpecLike): boolean {
 	return table !== undefined && getAliasIndex(table).familyIds.has(spec.id);
 }
 
+/**
+ * Re-point a stale collapsed spec whose `requestModelId` or routing still
+ * targets a retired wire id. Collapsed snapshots (bundled catalog, cache
+ * rows, previous-generation fallbacks) pass through collapsing untouched, so
+ * a hand-table routing fix would otherwise never reach them. Only retired
+ * targets are rewritten — presence-filtered routing decisions from live
+ * discovery stay authoritative for everything else. Per retired entry the
+ * table's route for that effort wins, then the off/first-live-member wire id,
+ * then the route is dropped (falls back to `requestModelId ?? id`). Returns
+ * `spec` by reference when nothing targets a retired id.
+ */
+function reconcileRetiredRouting<TSpec extends VariantSpecLike>(
+	spec: TSpec,
+	family: EffortVariantFamily,
+	retired: ReadonlySet<string>,
+): TSpec {
+	const routing = spec.thinking?.effortRouting;
+	const requestRetired = spec.requestModelId !== undefined && retired.has(spec.requestModelId);
+	let routingRetired = false;
+	if (routing !== undefined) {
+		for (const key in routing) {
+			const target = routing[key as Effort | "off"];
+			if (target !== undefined && retired.has(target)) {
+				routingRetired = true;
+				break;
+			}
+		}
+	}
+	if (!requestRetired && !routingRetired) return spec;
+
+	const offTarget = family.routing.off;
+	const fallbackWireId =
+		offTarget !== undefined && !retired.has(offTarget) ? offTarget : family.members.find(id => !retired.has(id));
+	const next: TSpec = { ...spec };
+	if (routingRetired && routing !== undefined) {
+		const nextRouting: Partial<Record<Effort | "off", string>> = {};
+		for (const key in routing) {
+			const effortKey = key as Effort | "off";
+			const target = routing[effortKey];
+			if (target === undefined) continue;
+			if (!retired.has(target)) {
+				nextRouting[effortKey] = target;
+				continue;
+			}
+			const tableTarget = family.routing[effortKey];
+			if (tableTarget !== undefined && !retired.has(tableTarget)) {
+				nextRouting[effortKey] = tableTarget;
+			} else if (fallbackWireId !== undefined) {
+				nextRouting[effortKey] = fallbackWireId;
+			}
+		}
+		next.thinking = { ...(spec.thinking as ThinkingConfig), effortRouting: nextRouting };
+	}
+	if (requestRetired) {
+		if (fallbackWireId !== undefined && fallbackWireId !== spec.id) {
+			next.requestModelId = fallbackWireId;
+		} else {
+			delete next.requestModelId;
+		}
+	}
+	return next;
+}
+
 /**
  * Collapse every family in `table` found in `specs`. Non-member specs pass
  * through verbatim (by reference), order preserved; the collapsed spec
@@ -302,13 +380,26 @@ export function collapseEffortVariants<TSpec extends VariantSpecLike>(
 	const familyIdBySpecId = new Map<string, string>();
 
 	for (const family of table.families) {
+		const retired =
+			family.retiredMembers !== undefined && family.retiredMembers.length > 0
+				? new Set(family.retiredMembers)
+				: undefined;
 		const existing = byId.get(family.id);
 		const existingCollapsed =
 			existing !== undefined &&
 			(existing.requestModelId !== undefined || existing.thinking?.effortRouting !== undefined);
+		const reconciled =
+			existing !== undefined && existingCollapsed && retired !== undefined
+				? reconcileRetiredRouting(existing, family, retired)
+				: existing;
 		const rawPresent = family.members.filter(id => byId.has(id) && !(id === family.id && existingCollapsed));
 		if (rawPresent.length === 0) {
 			// Inert (no members) or already collapsed (pass-through) — idempotence.
+			// A stale collapsed entry still gets retired routing re-pointed.
+			if (reconciled !== undefined && reconciled !== existing) {
+				familyIdBySpecId.set(family.id, family.id);
+				replacement.set(family.id, reconciled);
+			}
 			continue;
 		}
 
@@ -317,8 +408,8 @@ export function collapseEffortVariants<TSpec extends VariantSpecLike>(
 
 		if (existingCollapsed) {
 			// Mixed input: the collapsed entry (live truth) wins; stale raw
-			// members are deduped away.
-			replacement.set(family.id, existing);
+			// members are deduped away. Retired targets are re-pointed first.
+			replacement.set(family.id, reconciled as TSpec);
 			continue;
 		}
 
@@ -329,7 +420,7 @@ export function collapseEffortVariants<TSpec extends VariantSpecLike>(
 		let hasEffortRoute = false;
 		for (const effortKey in family.routing) {
 			const target = family.routing[effortKey as Effort | "off"];
-			if (target !== undefined && presentSet.has(target)) {
+			if (target !== undefined && presentSet.has(target) && !retired?.has(target)) {
 				routing[effortKey as Effort | "off"] = target;
 				hasRouting = true;
 				if (effortKey !== "off") hasEffortRoute = true;
@@ -356,12 +447,14 @@ export function collapseEffortVariants<TSpec extends VariantSpecLike>(
 			contextWindow: Math.max(...memberSpecs.map(spec => spec.contextWindow)),
 			maxTokens: Math.max(...memberSpecs.map(spec => spec.maxTokens)),
 		};
-		// The default wire id is the priority member; omit when it equals the
-		// logical id (bare/thinking pairs) — `resolveWireModelId` falls back.
-		if (rawPresent[0] === family.id) {
+		// The default wire id is the highest-priority live member; omit when it
+		// equals the logical id (bare/thinking pairs) — `resolveWireModelId`
+		// falls back. Retired members never become the default.
+		const defaultWireId = rawPresent.find(id => !retired?.has(id)) ?? rawPresent[0];
+		if (defaultWireId === family.id) {
 			delete collapsed.requestModelId;
 		} else {
-			collapsed.requestModelId = rawPresent[0] as string;
+			collapsed.requestModelId = defaultWireId as string;
 		}
 		if (reasoning) {
 			collapsed.thinking = thinking;