@oh-my-pi/pi-ai 14.0.2 → 14.0.4

package/CHANGELOG.md

@@ -2,6 +2,12 @@
 
 ## [Unreleased]
 
+## [14.0.3] - 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": "@oh-my-pi/pi-ai",
-	"version": "14.0.2",
+	"version": "14.0.4",
 	"description": "Unified LLM API with automatic model discovery and provider configuration",
 	"homepage": "https://github.com/can1357/oh-my-pi",
 	"author": "Can Boluk",
@@ -45,7 +45,7 @@
 		"@aws-sdk/client-bedrock-runtime": "^3",
 		"@bufbuild/protobuf": "^2.11",
 		"@google/genai": "^1.43",
-		"@oh-my-pi/pi-utils": "14.0.2",
+		"@oh-my-pi/pi-utils": "14.0.4",
 		"@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/amazon-bedrock.ts

@@ -19,7 +19,7 @@ import {
 	type ToolConfiguration,
 	ToolResultStatus,
 } from "@aws-sdk/client-bedrock-runtime";
-import { $env } from "@oh-my-pi/pi-utils";
+import { $env, $flag } from "@oh-my-pi/pi-utils";
 import { NodeHttpHandler } from "@smithy/node-http-handler";
 import type { Effort } from "../model-thinking";
 import { mapEffortToAnthropicAdaptiveEffort, requireSupportedEffort } from "../model-thinking";
@@ -102,14 +102,14 @@ export const streamBedrock: StreamFunction<"bedrock-converse-stream"> = (
 			config.region = config.region || $env.AWS_REGION || $env.AWS_DEFAULT_REGION;
 
 			// Support proxies that don't need authentication
-			if ($env.AWS_BEDROCK_SKIP_AUTH === "1") {
+			if ($flag("AWS_BEDROCK_SKIP_AUTH")) {
 				config.credentials = {
 					accessKeyId: "dummy-access-key",
 					secretAccessKey: "dummy-secret-key",
 				};
 			}
 
-			if ($env.AWS_BEDROCK_FORCE_HTTP1 === "1") {
+			if ($flag("AWS_BEDROCK_FORCE_HTTP1")) {
 				config.requestHandler = new NodeHttpHandler();
 			}
 		}
@@ -379,7 +379,7 @@ function supportsPromptCaching(model: Model<"bedrock-converse-stream">): boolean
 	if (id.includes("claude-haiku")) return true;
 	// Application inference profiles don't contain the model name in the ARN.
 	// Allow users to force cache points via environment variable.
-	if (typeof process !== "undefined" && process.env.AWS_BEDROCK_FORCE_CACHE === "1") return true;
+	if (typeof process !== "undefined" && $flag("AWS_BEDROCK_FORCE_CACHE")) return true;
 	return false;
 }
 

package/src/providers/openai-codex-responses.ts

@@ -1,5 +1,5 @@
 import * as os from "node:os";
-import { $env, abortableSleep, asRecord, logger, readSseJson, structuredCloneJSON } from "@oh-my-pi/pi-utils";
+import { $env, $flag, abortableSleep, asRecord, logger, readSseJson, structuredCloneJSON } from "@oh-my-pi/pi-utils";
 import type OpenAI from "openai";
 import type {
 	ResponseFunctionToolCall,
@@ -90,7 +90,7 @@ export function buildCodexSystemPrompt(args: { userSystemPrompt?: string }): Cod
 	};
 }
 
-const CODEX_DEBUG = $env.PI_CODEX_DEBUG === "1" || $env.PI_CODEX_DEBUG === "true";
+const CODEX_DEBUG = $flag("PI_CODEX_DEBUG");
 const CODEX_MAX_RETRIES = 5;
 const CODEX_RETRYABLE_STATUS = new Set([408, 429, 500, 502, 503, 504]);
 const CODEX_RETRY_DELAY_MS = 500;
@@ -197,7 +197,7 @@ function parseCodexPositiveInteger(value: string | undefined, fallback: number):
 }
 
 function isCodexWebSocketEnvEnabled(): boolean {
-	return $env.PI_CODEX_WEBSOCKET === "1" || $env.PI_CODEX_WEBSOCKET === "true";
+	return $flag("PI_CODEX_WEBSOCKET");
 }
 
 function getCodexWebSocketRetryBudget(): number {

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");
 

package/src/utils/schema/strict-mode.ts

@@ -1,3 +1,4 @@
+import { $flag } from "@oh-my-pi/pi-utils";
 import { type TUnsafe, Type } from "@sinclair/typebox";
 import { areJsonValuesEqual } from "./equality";
 import { COMBINATOR_KEYS, NON_STRUCTURAL_SCHEMA_KEYS } from "./fields";
@@ -26,7 +27,7 @@ export function StringEnum<const T extends readonly string[]>(
 	});
 }
 
-export const NO_STRICT = Bun.env.PI_NO_STRICT === "1";
+export const NO_STRICT = $flag("PI_NO_STRICT");
 
 const strictSchemaCache = new WeakMap<Record<string, unknown>, { schema: Record<string, unknown>; strict: boolean }>();
 function hasUnrepresentableStrictObjectMap(schema: Record<string, unknown>, seen?: WeakSet<object>): boolean {