@oh-my-pi/pi-catalog 16.5.0 → 16.5.1

package/CHANGELOG.md

@@ -2,6 +2,14 @@
 
 ## [Unreleased]
 
+## [16.5.1] - 2026-07-14
+
+### Fixed
+
+- Fixed reasoning effort mapping for Z.ai GLM-5.2 on the Anthropic messages endpoint to correctly use the two-tier scale (high, max) and emit output_config.effort.
+- Fixed an issue where stale cached model limits would override updated static catalog limits after a catalog fingerprint mismatch.
+- Fixed Cursor discovery to correctly preserve GetUsableModels max-mode metadata for premium models and invalidate stale cache entries.
+
 ## [16.4.3] - 2026-07-11
 
 ### Fixed

package/dist/types/types.d.ts

@@ -590,6 +590,8 @@ export interface Model<TApi extends Api = Api> {
     supportsTools?: boolean;
     /** GitLab Duo Workflow root namespace selected during catalog discovery. */
     gitlabDuoWorkflowRootNamespaceId?: string;
+    /** Cursor `max_mode` request flag returned by `GetUsableModels` for premium models that require max mode. */
+    cursorMaxMode?: boolean;
     cost: {
         input: number;
         output: number;

package/package.json

@@ -1,7 +1,7 @@
 {
 	"type": "module",
 	"name": "@oh-my-pi/pi-catalog",
-	"version": "16.5.0",
+	"version": "16.5.1",
 	"description": "Model catalog for omp: bundled model database, provider discovery descriptors, model identity, classification, and equivalence",
 	"homepage": "https://omp.sh",
 	"author": "Can Boluk",
@@ -34,12 +34,12 @@
 	},
 	"dependencies": {
 		"@bufbuild/protobuf": "^2.12.1",
-		"@oh-my-pi/pi-utils": "16.5.0",
+		"@oh-my-pi/pi-utils": "16.5.1",
 		"arktype": "2.2.3",
 		"zod": "^4"
 	},
 	"devDependencies": {
-		"@oh-my-pi/pi-ai": "16.5.0",
+		"@oh-my-pi/pi-ai": "16.5.1",
 		"@types/bun": "^1.3.14"
 	},
 	"engines": {

package/src/discovery/cursor.ts

@@ -35,6 +35,7 @@ const CursorModelDetailsSchema = type({
 	displayModelId: OptionalDisplayNameSchema.default(undefined),
 	aliases: CursorAliasesSchema.default(() => []),
 	"thinkingDetails?": "unknown",
+	maxMode: "boolean = false",
 });
 
 const CursorModelsInnerSchema = type("unknown[]");
@@ -290,6 +291,7 @@ function normalizeCursorModel(
 			name,
 			baseUrl: baseUrlOverride ?? reference.baseUrl,
 			reasoning,
+			cursorMaxMode: details.maxMode,
 		};
 	}
 	return {
@@ -303,6 +305,7 @@ function normalizeCursorModel(
 		cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
 		contextWindow: DEFAULT_CONTEXT_WINDOW,
 		maxTokens: DEFAULT_MAX_TOKENS,
+		cursorMaxMode: details.maxMode,
 	};
 }
 

package/src/model-manager.ts

@@ -152,8 +152,9 @@ export async function resolveProviderModels<TApi extends Api = Api, TModelsDevPa
 	const dynamicFetchSucceeded = fetchedDynamicModels !== null;
 	const cacheModels = dynamicFetchSucceeded
 		? []
-		: dropCachedModelIdsOnStaticMismatch(
+		: prepareCacheModelsForStaticMismatch(
 				normalizeModelList<TApi>(cache?.models ?? []),
+				staticModels,
 				cacheFingerprintMatches,
 				options.dropCachedModelIdsOnStaticMismatch,
 			);
@@ -188,8 +189,9 @@ export async function resolveProviderModels<TApi extends Api = Api, TModelsDevPa
 				collapseBuiltModelVariants(
 					mergeDynamicModels(
 						mergeModelSources(staticModels, modelsDevModels),
-						dropCachedModelIdsOnStaticMismatch(
+						prepareCacheModelsForStaticMismatch(
 							normalizeModelList<TApi>(latestCache?.models ?? cache?.models ?? []),
+							staticModels,
 							cacheFingerprintMatches,
 							options.dropCachedModelIdsOnStaticMismatch,
 						),
@@ -260,16 +262,29 @@ function shouldFetchRemoteSources(
 	return false;
 }
 
-function dropCachedModelIdsOnStaticMismatch<TApi extends Api>(
+function prepareCacheModelsForStaticMismatch<TApi extends Api>(
 	models: readonly Model<TApi>[],
+	staticModels: readonly Model<TApi>[],
 	cacheFingerprintMatches: boolean,
 	ids: readonly string[] | undefined,
 ): Model<TApi>[] {
-	if (cacheFingerprintMatches || ids === undefined || ids.length === 0 || models.length === 0) {
-		return models.length === 0 ? [] : [...models];
+	if (models.length === 0) {
+		return [];
+	}
+	if (cacheFingerprintMatches) {
+		return [...models];
+	}
+
+	const droppedIds = ids && ids.length > 0 ? new Set(ids) : undefined;
+	const staticIds = staticModels.length > 0 ? new Set(staticModels.map(model => model.id)) : undefined;
+	const sanitizedModels: Model<TApi>[] = [];
+	for (const model of models) {
+		if (droppedIds?.has(model.id)) {
+			continue;
+		}
+		sanitizedModels.push(staticIds?.has(model.id) ? { ...model, contextWindow: null, maxTokens: null } : model);
 	}
-	const droppedIds = new Set(ids);
-	return models.filter(model => !droppedIds.has(model.id));
+	return sanitizedModels;
 }
 
 function mergeModelSources<TApi extends Api>(...sources: readonly (readonly Model<TApi>[])[]): Model<TApi>[] {

package/src/model-thinking.ts

@@ -318,7 +318,7 @@ function getModelDefinedEfforts<TApi extends Api>(
 		}
 		if (
 			isZaiThinkingFormat(compat) ||
-			isUmansGlm52ReasoningEffortModel(spec) ||
+			isAnthropicMessagesGlm52ReasoningEffortModel(spec) ||
 			isOllamaCloudGlm52ReasoningEffortModel(spec) ||
 			spec.provider === "baseten"
 		) {
@@ -394,8 +394,12 @@ function isOllamaCloudGlm52ReasoningEffortModel<TApi extends Api>(spec: ModelSpe
 	return spec.api === "ollama-chat" && spec.provider === "ollama-cloud" && isGlm52ReasoningEffortModelId(spec.id);
 }
 
-function isUmansGlm52ReasoningEffortModel<TApi extends Api>(spec: ModelSpec<TApi>): boolean {
-	return spec.api === "anthropic-messages" && spec.provider === "umans" && isGlm52ReasoningEffortModelId(spec.id);
+function isAnthropicMessagesGlm52ReasoningEffortModel<TApi extends Api>(spec: ModelSpec<TApi>): boolean {
+	return (
+		spec.api === "anthropic-messages" &&
+		(spec.provider === "umans" || spec.provider === "zai") &&
+		isGlm52ReasoningEffortModelId(spec.id)
+	);
 }
 
 function isMinimaxReasoningModelOnAnthropicEndpoint<TApi extends Api>(spec: ModelSpec<TApi>): boolean {
@@ -617,7 +621,7 @@ function inferThinkingControlMode<TApi extends Api>(
 			if (isMinimaxReasoningModelOnAnthropicEndpoint(spec)) {
 				return "anthropic-adaptive";
 			}
-			if (isUmansGlm52ReasoningEffortModel(spec)) {
+			if (isAnthropicMessagesGlm52ReasoningEffortModel(spec)) {
 				return "anthropic-budget-effort";
 			}
 			if (parsedModel.family === "anthropic") {

package/src/models.json

@@ -88484,13 +88484,10 @@
 			"contextWindow": 1000000,
 			"maxTokens": 131072,
 			"thinking": {
-				"mode": "budget",
+				"mode": "anthropic-budget-effort",
 				"efforts": [
-					"minimal",
-					"low",
-					"medium",
 					"high",
-					"xhigh"
+					"max"
 				]
 			}
 		},

package/src/provider-models/special.ts

@@ -42,10 +42,13 @@ export interface CursorModelManagerConfig {
 	clientVersion?: string;
 }
 
+const CURSOR_CACHE_PROVIDER_ID = "cursor:max-mode-v2";
+
 export function cursorModelManagerOptions(config: CursorModelManagerConfig = {}): ModelManagerOptions<"cursor-agent"> {
 	const { apiKey, baseUrl, clientVersion } = config;
 	return {
 		providerId: "cursor",
+		cacheProviderId: CURSOR_CACHE_PROVIDER_ID,
 		...(apiKey
 			? {
 					fetchDynamicModels: async () => {

package/src/types.ts

@@ -718,6 +718,8 @@ export interface Model<TApi extends Api = Api> {
 	supportsTools?: boolean;
 	/** GitLab Duo Workflow root namespace selected during catalog discovery. */
 	gitlabDuoWorkflowRootNamespaceId?: string;
+	/** Cursor `max_mode` request flag returned by `GetUsableModels` for premium models that require max mode. */
+	cursorMaxMode?: boolean;
 	cost: {
 		input: number; // $/million tokens
 		output: number; // $/million tokens