feat(catalog): integrated baseten provider and updated model definitions

- Implement Baseten provider support with authentication and dynamic model discovery.
- Register Baseten in the model catalog and provider priority order.
- Expand model definitions with new DeepSeek, Kimi, NVIDIA, and Claude variants.
- Update model configurations, cost data, and provider-specific metadata.

packages/ai/CHANGELOG.md

@@ -2,6 +2,10 @@
 
 ## [Unreleased]
 
+### Added
+
+- Added support for Baseten as an AI provider
+
 ## [16.3.3] - 2026-07-02
 
 ### Added

packages/ai/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;

packages/ai/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,

packages/catalog/CHANGELOG.md

@@ -2,6 +2,20 @@
 
 ## [Unreleased]
 
+### Added
+
+- Added Baseten as a supported model provider
+- Added support for new models from Baseten, including DeepSeek V4 Pro and Kimi series
+- Added new Devin agent models: Claude 5 Fable variants
+- Added new Github Copilot models: Kimi K2.7 Code and MAI-Code-1-Flash
+- Added Poolside Laguna XS 2.1 models via Kilo and OpenRouter providers
+- Added support for Claude Fable 5 (Free) via Zenmux provider
+
+### Changed
+
+- Updated priority ordering to include Baseten
+- Updated pricing and limits for various existing models in the catalog
+
 ## [16.3.3] - 2026-07-02
 
 ### Fixed

packages/catalog/scripts/generate-models.ts

@@ -187,7 +187,11 @@ function applyGlobalModelsDevFallback(
 	const providerScopedKeys = new Set(modelsDevModels.map(model => `${model.provider}/${model.id}`));
 	const globalReferences = createGlobalModelsDevReferenceMap(modelsDevModels);
 	return models.map(model => {
-		if (providerScopedKeys.has(`${model.provider}/${model.id}`) || model.provider === "devin") {
+		if (
+			providerScopedKeys.has(`${model.provider}/${model.id}`) ||
+			model.provider === "devin" ||
+			model.provider === "baseten"
+		) {
 			return model;
 		}
 		const reference = globalReferences.get(model.id);

packages/catalog/src/hosts.ts

@@ -47,6 +47,7 @@ export const KNOWN_HOSTS = {
 	xai: { providers: ["xai"], urlMarkers: ["api.x.ai"] },
 	mistral: { providers: ["mistral"], urlMarkers: ["mistral.ai"] },
 	together: { providers: ["together"], urlMarkers: ["api.together.xyz"] },
+	baseten: { providers: ["baseten"], urlMarkers: ["baseten.co"] },
 	/** URL-only on purpose: the `fireworks`/`firepass` providers route per-model and not every model is Fireworks-shaped. */
 	fireworks: { urlMarkers: ["fireworks.ai"] },
 	groq: { providers: ["groq"], urlMarkers: ["api.groq.com"] },

packages/catalog/src/identity/priority.ts

@@ -20,6 +20,7 @@ const DEFAULT_MODEL_PROVIDER_ORDER = [
 	// High-quality aggregators / hosted inference providers.
 	"fireworks",
 	"cerebras",
+	"baseten",
 	"openrouter",
 	"aimlapi",
 	"together",

packages/catalog/src/provider-models/descriptors.ts

@@ -12,6 +12,7 @@ import {
 	aimlApiModelManagerOptions,
 	alibabaCodingPlanModelManagerOptions,
 	anthropicModelManagerOptions,
+	basetenModelManagerOptions,
 	cerebrasModelManagerOptions,
 	cloudflareAiGatewayModelManagerOptions,
 	coreWeaveModelManagerOptions,
@@ -73,6 +74,14 @@ export const CATALOG_PROVIDERS = [
 		createModelManagerOptions: (config: ModelManagerConfig) => alibabaCodingPlanModelManagerOptions(config),
 		catalogDiscovery: { label: "Alibaba Coding Plan" },
 	},
+	{
+		id: "baseten",
+		defaultModel: "moonshotai/Kimi-K2.7-Code",
+		envVars: ["BASETEN_API_KEY"],
+		createModelManagerOptions: (config: ModelManagerConfig) => basetenModelManagerOptions(config),
+		dynamicModelsAuthoritative: true,
+		catalogDiscovery: { label: "Baseten" },
+	},
 	{
 		id: "amazon-bedrock",
 		defaultModel: "us.anthropic.claude-opus-4-8",

packages/catalog/src/provider-models/openai-compat.ts

@@ -2551,6 +2551,103 @@ export function veniceModelManagerOptions(
 	};
 }
 
+// ---------------------------------------------------------------------------
+// 14.5 Baseten
+// ---------------------------------------------------------------------------
+
+export interface BasetenModelManagerConfig {
+	apiKey?: string;
+	baseUrl?: string;
+	fetch?: FetchImpl;
+}
+
+export function basetenModelManagerOptions(
+	config?: BasetenModelManagerConfig,
+): ModelManagerOptions<"openai-completions"> {
+	const apiKey = config?.apiKey;
+	const baseUrl = config?.baseUrl ?? "https://inference.baseten.co/v1";
+	const references = createBundledReferenceMap<"openai-completions">("baseten");
+	return {
+		providerId: "baseten",
+		dynamicModelsAuthoritative: true,
+		...(apiKey && {
+			fetchDynamicModels: () =>
+				fetchOpenAICompatibleModels({
+					api: "openai-completions",
+					provider: "baseten",
+					baseUrl,
+					apiKey,
+					mapModel: (entry, defaults) => {
+						const reference = references.get(defaults.id);
+						const raw = entry as Record<string, unknown> & {
+							supported_features?: unknown;
+							input_modalities?: unknown;
+							pricing?: Record<string, unknown>;
+						};
+						const features = Array.isArray(raw.supported_features) ? raw.supported_features : [];
+						const modalities = Array.isArray(raw.input_modalities) ? raw.input_modalities : [];
+
+						const isBasetenNativeReasoning =
+							defaults.id === "openai/gpt-oss-120b" ||
+							defaults.id === "deepseek-ai/DeepSeek-V4-Pro" ||
+							defaults.id === "zai-org/GLM-5.2";
+						const reasoning =
+							isBasetenNativeReasoning &&
+							(features.includes("reasoning") || features.includes("reasoning_effort"));
+						const supportsTools = features.includes("tools") ? undefined : false;
+						const vision = modalities.includes("image") || (reference?.input.includes("image") ?? false);
+
+						const pricing = raw.pricing ?? {};
+						const cost = {
+							input: toPositiveNumber(pricing.prompt, 0) * 1_000_000,
+							output: toPositiveNumber(pricing.completion, 0) * 1_000_000,
+							cacheRead: toPositiveNumber(pricing.input_cache_read, 0) * 1_000_000,
+							cacheWrite: 0,
+						};
+
+						const contextWindow = toPositiveNumber(
+							raw.context_length,
+							reference?.contextWindow ?? defaults.contextWindow,
+						);
+						const maxTokens = toPositiveNumber(
+							raw.max_completion_tokens,
+							reference?.maxTokens ?? defaults.maxTokens,
+						);
+
+						const baseModel = mapWithBundledReference(entry, defaults, reference);
+
+						const isEffortReasoning = defaults.id === "openai/gpt-oss-120b" || defaults.id === "zai-org/GLM-5.2";
+						const thinking = isEffortReasoning
+							? {
+									mode: "effort" as const,
+									efforts: [Effort.Minimal, Effort.Low, Effort.Medium, Effort.High, Effort.XHigh],
+									effortMap: {
+										minimal: "high",
+										low: "high",
+										medium: "high",
+										high: "high",
+										xhigh: "max",
+									},
+								}
+							: undefined;
+
+						return {
+							...baseModel,
+							reasoning,
+							input: vision ? ["text", "image"] : ["text"],
+							cost,
+							contextWindow,
+							maxTokens,
+							...(thinking ? { thinking } : {}),
+							...(supportsTools === false ? { supportsTools } : {}),
+						};
+					},
+					fetch: config?.fetch,
+				}),
+		}),
+	};
+}
+
 // ---------------------------------------------------------------------------
 // 15. Together
 // ---------------------------------------------------------------------------

packages/catalog/test/baseten-provider.test.ts

@@ -0,0 +1,97 @@
+import { describe, expect, test } from "bun:test";
+import { basetenModelManagerOptions } from "@oh-my-pi/pi-catalog/provider-models/openai-compat";
+import type { FetchImpl } from "@oh-my-pi/pi-catalog/types";
+
+describe("Baseten provider discovery", () => {
+	test("discovers Baseten models with custom metadata", async () => {
+		const calls: Array<{ url: string; authorization: string | null }> = [];
+		const fetchMock: FetchImpl = async (input: string | URL | Request, init?: RequestInit) => {
+			const headers = new Headers(init?.headers);
+			calls.push({
+				url: String(input),
+				authorization: headers.get("authorization"),
+			});
+			return new Response(
+				JSON.stringify({
+					data: [
+						{
+							id: "moonshotai/Kimi-K2.7-Code",
+							object: "model",
+							name: "Kimi K2.7 Code",
+							context_length: 262000,
+							max_completion_tokens: 262000,
+							supported_features: ["tools", "json_mode", "structured_outputs", "reasoning"],
+							input_modalities: ["text", "image"],
+							pricing: {
+								prompt: "0.00000095",
+								completion: "0.000004",
+								input_cache_read: "0.00000016",
+							},
+						},
+						{
+							id: "deepseek-ai/DeepSeek-V4-Pro",
+							object: "model",
+							name: "DeepSeek V4 Pro",
+							context_length: 262144,
+							max_completion_tokens: 262144,
+							supported_features: ["tools", "json_mode", "structured_outputs", "reasoning"],
+							input_modalities: ["text"],
+							pricing: {
+								prompt: "0.00000174",
+								completion: "0.00000348",
+								input_cache_read: "0.000000145",
+							},
+						},
+					],
+				}),
+				{ status: 200, headers: { "content-type": "application/json" } },
+			);
+		};
+
+		const options = basetenModelManagerOptions({ apiKey: "baseten-test-key", fetch: fetchMock });
+		const models = await options.fetchDynamicModels?.();
+
+		expect(calls).toEqual([
+			{
+				url: "https://inference.baseten.co/v1/models",
+				authorization: "Bearer baseten-test-key",
+			},
+		]);
+
+		const kimi = models?.find(model => model.id === "moonshotai/Kimi-K2.7-Code");
+		expect(kimi).toBeDefined();
+		expect(kimi).toMatchObject({
+			provider: "baseten",
+			api: "openai-completions",
+			name: "Kimi K2.7 Code",
+			reasoning: false,
+			input: ["text", "image"],
+			contextWindow: 262000,
+			maxTokens: 262000,
+			cost: {
+				input: 0.95,
+				output: 4,
+				cacheRead: 0.16,
+				cacheWrite: 0,
+			},
+		});
+
+		const deepseek = models?.find(model => model.id === "deepseek-ai/DeepSeek-V4-Pro");
+		expect(deepseek).toBeDefined();
+		expect(deepseek).toMatchObject({
+			provider: "baseten",
+			api: "openai-completions",
+			name: "DeepSeek V4 Pro",
+			reasoning: true,
+			input: ["text"],
+			contextWindow: 262144,
+			maxTokens: 262144,
+			cost: {
+				input: 1.74,
+				output: 3.48,
+				cacheRead: 0.145,
+				cacheWrite: 0,
+			},
+		});
+	});
+});