@f5xc-salesdemos/pi-ai 14.2.1 → 14.3.0

package/CHANGELOG.md

@@ -2,6 +2,12 @@
 
 ## [Unreleased]
 
+## [14.3.0] - 2026-04-09
+
+### Fixed
+
+- Fixed Ollama discovery cache normalization so cached models upgrade to the OpenAI Responses transport after the provider change
+
 ## [14.0.0] - 2026-04-08
 ### Breaking Changes
 
@@ -1988,4 +1994,4 @@ _Dedicated to Peter's shoulder ([@steipete](https://twitter.com/steipete))_
 
 ## [0.9.4] - 2025-11-26
 
-Initial release with multi-provider LLM support.
+Initial release with multi-provider LLM support.

package/package.json

@@ -1,7 +1,7 @@
 {
 	"type": "module",
 	"name": "@f5xc-salesdemos/pi-ai",
-	"version": "14.2.1",
+	"version": "14.3.0",
 	"description": "Unified LLM API with automatic model discovery and provider configuration",
 	"homepage": "https://github.com/f5xc-salesdemos/xcsh",
 	"author": "Can Boluk",
@@ -45,7 +45,7 @@
 		"@aws-sdk/client-bedrock-runtime": "^3",
 		"@bufbuild/protobuf": "^2.11",
 		"@google/genai": "^1.43",
-		"@f5xc-salesdemos/pi-utils": "14.2.1",
+		"@f5xc-salesdemos/pi-utils": "14.3.0",
 		"@sinclair/typebox": "^0.34",
 		"@smithy/node-http-handler": "^4.4",
 		"ajv": "^8.18",

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

@@ -225,7 +225,7 @@ function toOllamaNativeBaseUrl(baseUrl: string): string {
 	return baseUrl.endsWith("/v1") ? baseUrl.slice(0, -3) : baseUrl;
 }
 
-async function fetchOllamaNativeModels(baseUrl: string): Promise<Model<"openai-completions">[] | null> {
+async function fetchOllamaNativeModels(baseUrl: string): Promise<Model<"openai-responses">[] | null> {
 	const nativeBaseUrl = toOllamaNativeBaseUrl(baseUrl);
 	let response: Response;
 	try {
@@ -241,7 +241,7 @@ async function fetchOllamaNativeModels(baseUrl: string): Promise<Model<"openai-c
 	}
 	const payload = (await response.json()) as { models?: Array<{ name?: string; model?: string }> };
 	const entries = payload.models ?? [];
-	const models: Model<"openai-completions">[] = [];
+	const models: Model<"openai-responses">[] = [];
 	for (const entry of entries) {
 		const id = entry.model ?? entry.name;
 		if (!id) {
@@ -250,7 +250,7 @@ async function fetchOllamaNativeModels(baseUrl: string): Promise<Model<"openai-c
 		models.push({
 			id,
 			name: entry.name ?? id,
-			api: "openai-completions",
+			api: "openai-responses",
 			provider: "ollama",
 			baseUrl,
 			reasoning: false,
@@ -602,19 +602,15 @@ export interface OllamaModelManagerConfig {
 	baseUrl?: string;
 }
 
-export function ollamaModelManagerOptions(
-	config?: OllamaModelManagerConfig,
-): ModelManagerOptions<"openai-completions"> {
+export function ollamaModelManagerOptions(config?: OllamaModelManagerConfig): ModelManagerOptions<"openai-responses"> {
 	const apiKey = config?.apiKey;
 	const baseUrl = normalizeOllamaBaseUrl(config?.baseUrl);
-	const references = createBundledReferenceMap<"openai-completions">(
-		"ollama" as Parameters<typeof getBundledModels>[0],
-	);
+	const references = createBundledReferenceMap<"openai-responses">("ollama" as Parameters<typeof getBundledModels>[0]);
 	return {
 		providerId: "ollama",
 		fetchDynamicModels: async () => {
 			const openAiCompatible = await fetchOpenAICompatibleModels({
-				api: "openai-completions",
+				api: "openai-responses",
 				provider: "ollama",
 				baseUrl,
 				apiKey,

package/src/providers/openai-completions-compat.ts

@@ -49,7 +49,7 @@ export function detectOpenAICompat(model: Model<"openai-completions">, resolvedB
 
 	const isCerebras = provider === "cerebras" || baseUrl.includes("cerebras.ai");
 	const isZai = provider === "zai" || baseUrl.includes("api.z.ai");
-	const isKimiModel = model.id.includes("moonshotai/kimi");
+	const isKimiModel = model.id.includes("moonshotai/kimi") || /^kimi[-.]/i.test(model.id);
 	const isAlibaba = provider === "alibaba-coding-plan" || baseUrl.includes("dashscope");
 	const isQwen = model.id.toLowerCase().includes("qwen");