@oh-my-pi/pi-ai 16.3.3 → 16.3.4

package/CHANGELOG.md

@@ -2,6 +2,20 @@
 
 ## [Unreleased]
 
+## [16.3.4] - 2026-07-03
+
+### Added
+
+- Added support for Baseten as an AI provider
+
+### Changed
+
+- Improved Claude usage reliability by removing proactive hard-blocking for Fable and Mythos tiers
+
+### Fixed
+
+- Fixed Anthropic OAuth account rotation to exclude unreliable model-scoped Fable/Mythos weekly caps from proactive hard-blocking, ensuring they act only as ranking priority hints while still allowing reactive 429-fallback to rotate and reach serviceable siblings.
+
 ## [16.3.3] - 2026-07-02
 
 ### Added

package/dist/types/registry/baseten.d.ts

@@ -0,0 +1,7 @@
+import type { OAuthLoginCallbacks } from "./oauth/types";
+export declare const loginBaseten: (options: import("./oauth").OAuthController) => Promise<string>;
+export declare const basetenProvider: {
+    readonly id: "baseten";
+    readonly name: "Baseten";
+    readonly login: (cb: OAuthLoginCallbacks) => Promise<string>;
+};

package/dist/types/registry/registry.d.ts

@@ -29,6 +29,10 @@ declare const ALL: ({
 } | {
     readonly id: "azure";
     readonly name: "Azure OpenAI";
+} | {
+    readonly id: "baseten";
+    readonly name: "Baseten";
+    readonly login: (cb: import("./oauth").OAuthLoginCallbacks) => Promise<string>;
 } | {
     readonly id: "cerebras";
     readonly name: "Cerebras";

package/package.json

@@ -1,7 +1,7 @@
 {
 	"type": "module",
 	"name": "@oh-my-pi/pi-ai",
-	"version": "16.3.3",
+	"version": "16.3.4",
 	"description": "Unified LLM API with automatic model discovery and provider configuration",
 	"homepage": "https://omp.sh",
 	"author": "Can Boluk",
@@ -38,9 +38,9 @@
 	},
 	"dependencies": {
 		"@bufbuild/protobuf": "^2.12.0",
-		"@oh-my-pi/pi-catalog": "16.3.3",
-		"@oh-my-pi/pi-utils": "16.3.3",
-		"@oh-my-pi/pi-wire": "16.3.3",
+		"@oh-my-pi/pi-catalog": "16.3.4",
+		"@oh-my-pi/pi-utils": "16.3.4",
+		"@oh-my-pi/pi-wire": "16.3.4",
 		"arktype": "^2.2.0",
 		"zod": "^4"
 	},

package/src/registry/baseten.ts

@@ -0,0 +1,22 @@
+import { createApiKeyLogin } from "./api-key-login";
+import type { OAuthLoginCallbacks } from "./oauth/types";
+import type { ProviderDefinition } from "./types";
+
+export const loginBaseten = createApiKeyLogin({
+	providerLabel: "Baseten",
+	authUrl: "https://app.baseten.co/settings/api_keys",
+	instructions: "Copy your API key from the Baseten dashboard",
+	promptMessage: "Paste your Baseten API key",
+	placeholder: "bt_...",
+	validation: {
+		kind: "models-endpoint",
+		provider: "Baseten",
+		modelsUrl: "https://inference.baseten.co/v1/models",
+	},
+});
+
+export const basetenProvider = {
+	id: "baseten",
+	name: "Baseten",
+	login: (cb: OAuthLoginCallbacks) => loginBaseten(cb),
+} as const satisfies ProviderDefinition;

package/src/registry/registry.ts

@@ -4,6 +4,7 @@ import { alibabaCodingPlanProvider } from "./alibaba-coding-plan";
 import { amazonBedrockProvider } from "./amazon-bedrock";
 import { anthropicProvider } from "./anthropic";
 import { azureProvider } from "./azure";
+import { basetenProvider } from "./baseten";
 import { cerebrasProvider } from "./cerebras";
 import { cloudflareAiGatewayProvider } from "./cloudflare-ai-gateway";
 import { coreWeaveProvider } from "./coreweave";
@@ -105,6 +106,7 @@ const ALL = [
 	deepseekProvider,
 	moonshotProvider,
 	cerebrasProvider,
+	basetenProvider,
 	fireworksProvider,
 	togetherProvider,
 	nvidiaProvider,

package/src/usage/claude.ts

@@ -614,6 +614,23 @@ function scopeClaudeLimitsForModel(report: UsageReport, context: CredentialRanki
 	);
 }
 
+/**
+ * Exclude Fable and Mythos tier weekly caps from proactive hard-blocking
+ * (gating) because they are notoriously unreliable (they report 100% exhausted
+ * while the account can still serve requests). They stay available for ranking
+ * pressure in findWindowLimits via scopeClaudeLimitsForModel.
+ */
+function scopeClaudeLimitsForModelHardBlock(
+	report: UsageReport,
+	context: CredentialRankingContext | undefined,
+): UsageLimit[] {
+	const kind = getClaudeModelKind(context);
+	const excludeHardBlock = kind === "fable" || kind === "mythos";
+	return report.limits.filter(
+		limit => limit.scope.shared === true || (kind !== undefined && limit.scope.tier === kind && !excludeHardBlock),
+	);
+}
+
 function rankingUsedFraction(limit: UsageLimit): number {
 	const fraction = resolveUsedFraction(limit);
 	if (typeof fraction !== "number" || !Number.isFinite(fraction)) return 0.5;
@@ -664,7 +681,7 @@ export const claudeRankingStrategy: CredentialRankingStrategy = {
 		const secondary = findClaudeSecondaryLimit(report, context);
 		return { primary, secondary };
 	},
-	scopeLimits: scopeClaudeLimitsForModel,
+	scopeLimits: scopeClaudeLimitsForModelHardBlock,
 	/**
 	 * Fable/Mythos usage-limit errors map to tier-local weekly counters. Scope
 	 * reactive backoff blocks for those tiers, mirroring the per-counter