@oh-my-pi/pi-catalog 16.3.8 → 16.3.10

package/CHANGELOG.md

@@ -2,6 +2,18 @@
 
 ## [Unreleased]
 
+## [16.3.10] - 2026-07-06
+
+### Fixed
+
+- Fixed LiteLLM rich discovery to ignore unusable sentinel placeholders and continue to `/v2/model/info` for real models. ([#4655](https://github.com/can1357/oh-my-pi/issues/4655))
+
+## [16.3.9] - 2026-07-06
+
+### Fixed
+
+- Fixed compatibility with OpenCode Go DeepSeek V4 models by sending max_tokens instead of max_completion_tokens to match the provider's API requirements.
+
 ## [16.3.7] - 2026-07-05
 
 ### Fixed

package/package.json

@@ -1,7 +1,7 @@
 {
 	"type": "module",
 	"name": "@oh-my-pi/pi-catalog",
-	"version": "16.3.8",
+	"version": "16.3.10",
 	"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": "16.3.8",
+		"@oh-my-pi/pi-utils": "16.3.10",
 		"arktype": "^2.2.0",
 		"zod": "^4"
 	},
 	"devDependencies": {
-		"@oh-my-pi/pi-ai": "16.3.8",
+		"@oh-my-pi/pi-ai": "16.3.10",
 		"@types/bun": "^1.3.14"
 	},
 	"engines": {

package/src/models.json

@@ -60395,6 +60395,7 @@
 			"maxTokens": 384000,
 			"compat": {
 				"supportsToolChoice": false,
+				"maxTokensField": "max_tokens",
 				"reasoningContentField": "reasoning_content",
 				"requiresReasoningContentForToolCalls": true
 			},
@@ -60436,6 +60437,7 @@
 			"maxTokens": 384000,
 			"compat": {
 				"supportsToolChoice": false,
+				"maxTokensField": "max_tokens",
 				"reasoningContentField": "reasoning_content",
 				"requiresReasoningContentForToolCalls": true
 			},
@@ -71921,9 +71923,9 @@
 				"text"
 			],
 			"cost": {
-				"input": 0.5740000000000001,
-				"output": 1.804,
-				"cacheRead": 0.1066,
+				"input": 0.56,
+				"output": 1.76,
+				"cacheRead": 0.10400000000000001,
 				"cacheWrite": 0
 			},
 			"contextWindow": 1048576,
@@ -72143,7 +72145,7 @@
 	"synthetic": {
 		"hf:MiniMaxAI/MiniMax-M3": {
 			"id": "hf:MiniMaxAI/MiniMax-M3",
-			"name": "MiniMaxAI/MiniMax-M3",
+			"name": "MiniMax-M3",
 			"api": "openai-completions",
 			"provider": "synthetic",
 			"baseUrl": "https://api.synthetic.new/openai/v1",
@@ -72158,7 +72160,7 @@
 				"cacheRead": 0.6,
 				"cacheWrite": 0
 			},
-			"contextWindow": 262144,
+			"contextWindow": 524288,
 			"maxTokens": 65536,
 			"thinking": {
 				"mode": "effort",
@@ -72173,7 +72175,7 @@
 		},
 		"hf:moonshotai/Kimi-K2.7-Code": {
 			"id": "hf:moonshotai/Kimi-K2.7-Code",
-			"name": "moonshotai/Kimi-K2.7-Code",
+			"name": "Kimi K2.7 Code",
 			"api": "openai-completions",
 			"provider": "synthetic",
 			"baseUrl": "https://api.synthetic.new/openai/v1",
@@ -72203,7 +72205,7 @@
 		},
 		"hf:nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4": {
 			"id": "hf:nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4",
-			"name": "nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4",
+			"name": "Nemotron 3 Super 120B A12B",
 			"api": "openai-completions",
 			"provider": "synthetic",
 			"baseUrl": "https://api.synthetic.new/openai/v1",
@@ -72232,7 +72234,7 @@
 		},
 		"hf:openai/gpt-oss-120b": {
 			"id": "hf:openai/gpt-oss-120b",
-			"name": "openai/gpt-oss-120b",
+			"name": "GPT OSS 120B",
 			"api": "openai-completions",
 			"provider": "synthetic",
 			"baseUrl": "https://api.synthetic.new/openai/v1",
@@ -72259,7 +72261,7 @@
 		},
 		"hf:Qwen/Qwen3.6-27B": {
 			"id": "hf:Qwen/Qwen3.6-27B",
-			"name": "Qwen/Qwen3.6-27B",
+			"name": "Qwen3.6 27B",
 			"api": "openai-completions",
 			"provider": "synthetic",
 			"baseUrl": "https://api.synthetic.new/openai/v1",
@@ -72288,7 +72290,7 @@
 		},
 		"hf:zai-org/GLM-4.7-Flash": {
 			"id": "hf:zai-org/GLM-4.7-Flash",
-			"name": "zai-org/GLM-4.7-Flash",
+			"name": "GLM-4.7-Flash",
 			"api": "openai-completions",
 			"provider": "synthetic",
 			"baseUrl": "https://api.synthetic.new/openai/v1",
@@ -72317,7 +72319,7 @@
 		},
 		"hf:zai-org/GLM-5.2": {
 			"id": "hf:zai-org/GLM-5.2",
-			"name": "zai-org/GLM-5.2",
+			"name": "GLM-5.2",
 			"api": "openai-completions",
 			"provider": "synthetic",
 			"baseUrl": "https://api.synthetic.new/openai/v1",
@@ -83217,6 +83219,14 @@
 			},
 			"contextWindow": 2000000,
 			"maxTokens": 2000000,
+			"compat": {
+				"reasoningEffortMap": {
+					"minimal": "low"
+				},
+				"includeEncryptedReasoning": false,
+				"filterReasoningHistory": true,
+				"omitReasoningEffort": false
+			},
 			"thinking": {
 				"mode": "effort",
 				"efforts": [
@@ -83229,14 +83239,6 @@
 				"effortMap": {
 					"minimal": "low"
 				}
-			},
-			"compat": {
-				"reasoningEffortMap": {
-					"minimal": "low"
-				},
-				"includeEncryptedReasoning": false,
-				"filterReasoningHistory": true,
-				"omitReasoningEffort": false
 			}
 		},
 		"grok-4.3": {
@@ -83258,6 +83260,14 @@
 			},
 			"contextWindow": 1000000,
 			"maxTokens": 1000000,
+			"compat": {
+				"reasoningEffortMap": {
+					"minimal": "low"
+				},
+				"includeEncryptedReasoning": false,
+				"filterReasoningHistory": true,
+				"omitReasoningEffort": false
+			},
 			"thinking": {
 				"mode": "effort",
 				"efforts": [
@@ -83270,14 +83280,6 @@
 				"effortMap": {
 					"minimal": "low"
 				}
-			},
-			"compat": {
-				"reasoningEffortMap": {
-					"minimal": "low"
-				},
-				"includeEncryptedReasoning": false,
-				"filterReasoningHistory": true,
-				"omitReasoningEffort": false
 			}
 		},
 		"grok-build": {

package/src/provider-models/openai-compat.ts

@@ -3037,6 +3037,11 @@ const LITELLM_RICH_ENDPOINTS = ["/model_group/info", "/v2/model/info", "/model/i
 export const OPENAI_COMPAT_DISCOVERY_DEFAULT_CONTEXT_WINDOW = 128_000;
 export const OPENAI_COMPAT_DISCOVERY_DEFAULT_MAX_TOKENS = 32_768;
 const UNKNOWN_PROXY_COST = { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 } as const;
+const LITELLM_UNUSABLE_SENTINEL_IDS: Record<string, true> = {
+	"all-team-models": true,
+	"all-proxy-models": true,
+	"no-default-models": true,
+};
 
 export function normalizeLiteLLMManagementBaseUrl(baseUrl: string): string {
 	const trimmed = baseUrl.trim().replace(/\/+$/g, "");
@@ -3142,11 +3147,59 @@ function getSupportedOpenAIParams(entry: LiteLLMRichModelEntry): string[] | unde
 	return value.flatMap(item => (typeof item === "string" ? [item] : []));
 }
 
+function isLiteLLMUnusableSentinelPlaceholder(entry: LiteLLMRichModelEntry): boolean {
+	const modelGroup = toNonEmptyString(entry.model_group);
+	const id = toNonEmptyString(entry.id);
+	if (
+		(modelGroup === undefined || LITELLM_UNUSABLE_SENTINEL_IDS[modelGroup] !== true) &&
+		(id === undefined || LITELLM_UNUSABLE_SENTINEL_IDS[id] !== true)
+	) {
+		return false;
+	}
+	const providers = entry.providers;
+	if (providers !== undefined && (!Array.isArray(providers) || providers.length > 0)) {
+		return false;
+	}
+	const modelName = toNonEmptyString(entry.model_name);
+	if (modelName && LITELLM_UNUSABLE_SENTINEL_IDS[modelName] !== true) {
+		return false;
+	}
+	if (id && LITELLM_UNUSABLE_SENTINEL_IDS[id] !== true) {
+		return false;
+	}
+	const backendModel = toNonEmptyString(getLiteLLMParams(entry)?.model);
+	if (backendModel && LITELLM_UNUSABLE_SENTINEL_IDS[backendModel] !== true) {
+		return false;
+	}
+	if (
+		toPositiveNumber(getLiteLLMMetadataValue(entry, "max_input_tokens"), null) !== null ||
+		toPositiveNumber(getLiteLLMMetadataValue(entry, "max_output_tokens"), null) !== null
+	) {
+		return false;
+	}
+	if (
+		getLiteLLMMetadataValue(entry, "supports_vision") === true ||
+		getLiteLLMMetadataValue(entry, "supports_reasoning") === true ||
+		getLiteLLMMetadataValue(entry, "supports_function_calling") === true ||
+		getLiteLLMMetadataValue(entry, "supports_tools") === true
+	) {
+		return false;
+	}
+	const supportedOpenAIParams = getSupportedOpenAIParams(entry);
+	if (supportedOpenAIParams && supportedOpenAIParams.length > 0) {
+		return false;
+	}
+	return true;
+}
+
 function mapLiteLLMRichEntry<TApi extends Api>(
 	entry: LiteLLMRichModelEntry,
 	options: FetchLiteLLMRichModelsOptions<TApi>,
 	runtimeBaseUrl: string,
 ): ModelSpec<TApi> | null {
+	if (isLiteLLMUnusableSentinelPlaceholder(entry)) {
+		return null;
+	}
 	const id = getLiteLLMRichModelId(entry);
 	if (!id) {
 		return null;
@@ -3286,11 +3339,11 @@ export function litellmModelManagerOptions(
 	const baseUrl = config?.baseUrl ?? Bun.env.LITELLM_BASE_URL ?? "http://localhost:4000/v1";
 	return {
 		providerId: "litellm",
-		// rich-v2 invalidates rows cached before reseller usage-suffix stripping
-		// (stale display names like `MiniMax-M3 (3x usage)`); bump the version
-		// whenever the mappers below change, or warm authoritative caches keep
-		// serving pre-change rows for the full TTL.
-		cacheProviderId: `litellm:rich-v2:${Bun.hash(baseUrl).toString(36)}`,
+		// rich-v3 invalidates rows cached before reseller usage-suffix stripping
+		// and placeholder-only `all-team-models` filtering; bump the version whenever
+		// the mappers below change, or warm authoritative caches keep serving
+		// pre-change rows for the full TTL.
+		cacheProviderId: `litellm:rich-v3:${Bun.hash(baseUrl).toString(36)}`,
 		// litellm is a local-only proxy and is never bundled in models.json (that
 		// would leak the machine's localhost catalog). Prefer the proxy's richer
 		// management metadata, then fall back to /v1/models and enrich bare ids