@oh-my-pi/pi-catalog 16.2.2 → 16.2.3

package/CHANGELOG.md

@@ -2,6 +2,18 @@
 
 ## [Unreleased]
 
+## [16.2.3] - 2026-06-28
+
+### Added
+
+- Added support and configuration parameters for V2 streaming compaction in RemoteCompactionConfig, catalog types, and model/provider metadata.
+
+### Changed
+
+- Enabled automatic content markup healing for all OpenAI-compatible streaming models
+- Updated pricing and context window limits for several catalog models.
+- Disabled reasoning capability for multiple providers in the catalog.
+
 ## [16.2.2] - 2026-06-27
 
 ### Removed

package/dist/types/types.d.ts

@@ -532,8 +532,14 @@ export interface RemoteCompactionConfig<TApi extends Api = Api> {
     enabled?: boolean;
     /** Adapter family used by the configured compaction endpoint. */
     api?: TApi;
-    /** Absolute compact endpoint URL; when omitted, the adapter derives it from the model base URL. */
+    /** Absolute V1 compact endpoint URL; when omitted, the adapter derives it from the model base URL. */
     endpoint?: string;
+    /** Enables Responses-stream V2 compaction for models verified to support `compaction_trigger`. */
+    v2StreamingEnabled?: boolean;
+    /** Absolute Responses-stream endpoint URL for V2 compaction; overrides `streamingEndpoint`. */
+    v2Endpoint?: string;
+    /** Absolute provider streaming endpoint URL used by V2 compaction when no dedicated endpoint is set. */
+    streamingEndpoint?: string;
     /** Model id sent to the compaction endpoint when it differs from the active model id. */
     model?: string;
 }

package/package.json

@@ -1,7 +1,7 @@
 {
 	"type": "module",
 	"name": "@oh-my-pi/pi-catalog",
-	"version": "16.2.2",
+	"version": "16.2.3",
 	"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": "^2.12.0",
-		"@oh-my-pi/pi-utils": "16.2.2",
+		"@oh-my-pi/pi-utils": "16.2.3",
 		"arktype": "^2.2.0",
 		"zod": "^4"
 	},
 	"devDependencies": {
-		"@oh-my-pi/pi-ai": "16.2.2",
+		"@oh-my-pi/pi-ai": "16.2.3",
 		"@types/bun": "^1.3.14"
 	},
 	"engines": {

package/src/compat/openai.ts

@@ -98,20 +98,21 @@ function resolveReasoningDisableMode(
 	}
 }
 
-function detectStreamMarkupHealingPattern(
-	provider: string,
-	modelId: string,
-): OpenAIStreamMarkupHealingPattern | undefined {
-	if (MINIMAX_PROVIDER_OR_ID_PATTERN.test(provider) || MINIMAX_PROVIDER_OR_ID_PATTERN.test(modelId)) {
-		return "thinking";
-	}
+/**
+ * Pick the leaked-markup healer for an OpenAI-compatible visible-text stream.
+ * Kimi chat-template tokens and DeepSeek DSML envelopes need their dedicated
+ * tool-call grammars; every other model defaults to `"thinking"`. All patterns
+ * run the generic thinking healer, so leaked reasoning idioms (e.g. a Gemini
+ * ` ```thinking ` fence on OpenRouter) are always recovered from `delta.content`.
+ */
+function detectStreamMarkupHealingPattern(provider: string, modelId: string): OpenAIStreamMarkupHealingPattern {
 	if (provider === "kimi-code" || provider === "moonshot" || /kimi[-/_.]?k2/i.test(modelId)) {
 		return "kimi";
 	}
 	if (isDeepseekModelIdOrName(modelId) && DSML_HEALING_PROVIDERS.has(provider)) {
 		return "dsml";
 	}
-	return undefined;
+	return "thinking";
 }
 
 /**

package/src/models.json

@@ -31303,7 +31303,7 @@
 			"api": "openai-completions",
 			"provider": "kilo",
 			"baseUrl": "https://api.kilo.ai/api/gateway",
-			"reasoning": true,
+			"reasoning": false,
 			"input": [
 				"text"
 			],
@@ -31314,16 +31314,7 @@
 				"cacheWrite": 0
 			},
 			"contextWindow": 262144,
-			"maxTokens": 52429,
-			"thinking": {
-				"mode": "effort",
-				"efforts": [
-					"minimal",
-					"low",
-					"medium",
-					"high"
-				]
-			}
+			"maxTokens": 52429
 		},
 		"qwen/qwen3-235b-a22b-thinking-2507": {
 			"id": "qwen/qwen3-235b-a22b-thinking-2507",
@@ -68464,7 +68455,7 @@
 			"api": "openrouter",
 			"baseUrl": "https://openrouter.ai/api/v1",
 			"provider": "openrouter",
-			"reasoning": true,
+			"reasoning": false,
 			"input": [
 				"text"
 			],
@@ -68475,16 +68466,7 @@
 				"cacheWrite": 0
 			},
 			"contextWindow": 262144,
-			"maxTokens": 16384,
-			"thinking": {
-				"mode": "effort",
-				"efforts": [
-					"minimal",
-					"low",
-					"medium",
-					"high"
-				]
-			}
+			"maxTokens": 16384
 		},
 		"qwen/qwen3-235b-a22b-thinking-2507": {
 			"id": "qwen/qwen3-235b-a22b-thinking-2507",
@@ -76863,13 +76845,13 @@
 				"text"
 			],
 			"cost": {
-				"input": 0.56,
-				"output": 1.68,
+				"input": 0.6,
+				"output": 1.7,
 				"cacheRead": 0.28,
 				"cacheWrite": 0
 			},
-			"contextWindow": 163840,
-			"maxTokens": 8192,
+			"contextWindow": 128000,
+			"maxTokens": 128000,
 			"thinking": {
 				"mode": "budget",
 				"efforts": [
@@ -78430,13 +78412,13 @@
 				"text"
 			],
 			"cost": {
-				"input": 0.6,
-				"output": 2.5,
-				"cacheRead": 0.15,
+				"input": 0.47,
+				"output": 2,
+				"cacheRead": 0.14100000000000001,
 				"cacheWrite": 0
 			},
-			"contextWindow": 262114,
-			"maxTokens": 262114,
+			"contextWindow": 216144,
+			"maxTokens": 216144,
 			"thinking": {
 				"mode": "budget",
 				"efforts": [
@@ -86461,7 +86443,7 @@
 			"api": "openai-completions",
 			"provider": "zenmux",
 			"baseUrl": "https://zenmux.ai/api/v1",
-			"reasoning": true,
+			"reasoning": false,
 			"input": [
 				"text"
 			],
@@ -86472,16 +86454,7 @@
 				"cacheWrite": 0
 			},
 			"contextWindow": 256000,
-			"maxTokens": 16384,
-			"thinking": {
-				"mode": "effort",
-				"efforts": [
-					"minimal",
-					"low",
-					"medium",
-					"high"
-				]
-			}
+			"maxTokens": 16384
 		},
 		"qwen/qwen3-235b-a22b-thinking-2507": {
 			"id": "qwen/qwen3-235b-a22b-thinking-2507",

package/src/types.ts

@@ -659,8 +659,14 @@ export interface RemoteCompactionConfig<TApi extends Api = Api> {
 	enabled?: boolean;
 	/** Adapter family used by the configured compaction endpoint. */
 	api?: TApi;
-	/** Absolute compact endpoint URL; when omitted, the adapter derives it from the model base URL. */
+	/** Absolute V1 compact endpoint URL; when omitted, the adapter derives it from the model base URL. */
 	endpoint?: string;
+	/** Enables Responses-stream V2 compaction for models verified to support `compaction_trigger`. */
+	v2StreamingEnabled?: boolean;
+	/** Absolute Responses-stream endpoint URL for V2 compaction; overrides `streamingEndpoint`. */
+	v2Endpoint?: string;
+	/** Absolute provider streaming endpoint URL used by V2 compaction when no dedicated endpoint is set. */
+	streamingEndpoint?: string;
 	/** Model id sent to the compaction endpoint when it differs from the active model id. */
 	model?: string;
 }