refactor(ai): Update LLM implementations to use Model objects

- LLM constructors now take Model objects instead of string IDs
- Added provider field to AssistantMessage interface
- Updated getModel function with type-safe model ID autocomplete
- Fixed Anthropic model ID mapping for proper API aliases
- Added baseUrl to Model interface for provider-specific endpoints
- Updated all tests to use getModel for model instantiation
- Removed deprecated models.json in favor of generated models

packages/ai/src/providers/anthropic.ts

@@ -11,6 +11,7 @@ import type {
 	LLM,
 	LLMOptions,
 	Message,
+	Model,
 	StopReason,
 	TokenUsage,
 	ToolCall,
@@ -26,10 +27,10 @@ export interface AnthropicLLMOptions extends LLMOptions {
 
 export class AnthropicLLM implements LLM<AnthropicLLMOptions> {
 	private client: Anthropic;
-	private model: string;
+	private modelInfo: Model;
 	private isOAuthToken: boolean = false;
 
-	constructor(model: string, apiKey?: string, baseUrl?: string) {
+	constructor(model: Model, apiKey?: string) {
 		if (!apiKey) {
 			if (!process.env.ANTHROPIC_API_KEY) {
 				throw new Error(
@@ -45,13 +46,17 @@ export class AnthropicLLM implements LLM<AnthropicLLMOptions> {
 			};
 
 			process.env.ANTHROPIC_API_KEY = undefined;
-			this.client = new Anthropic({ apiKey: null, authToken: apiKey, baseURL: baseUrl, defaultHeaders });
+			this.client = new Anthropic({ apiKey: null, authToken: apiKey, baseURL: model.baseUrl, defaultHeaders });
 			this.isOAuthToken = true;
 		} else {
-			this.client = new Anthropic({ apiKey, baseURL: baseUrl });
+			this.client = new Anthropic({ apiKey, baseURL: model.baseUrl });
 			this.isOAuthToken = false;
 		}
-		this.model = model;
+		this.modelInfo = model;
+	}
+
+	getModel(): Model {
+		return this.modelInfo;
 	}
 
 	async complete(context: Context, options?: AnthropicLLMOptions): Promise<AssistantMessage> {
@@ -59,7 +64,7 @@ export class AnthropicLLM implements LLM<AnthropicLLMOptions> {
 			const messages = this.convertMessages(context.messages);
 
 			const params: MessageCreateParamsStreaming = {
-				model: this.model,
+				model: this.modelInfo.id,
 				messages,
 				max_tokens: options?.maxTokens || 4096,
 				stream: true,
@@ -97,7 +102,8 @@ export class AnthropicLLM implements LLM<AnthropicLLMOptions> {
 				params.tools = this.convertTools(context.tools);
 			}
 
-			if (options?.thinking?.enabled) {
+			// Only enable thinking if the model supports it
+			if (options?.thinking?.enabled && this.modelInfo.reasoning) {
 				params.thinking = {
 					type: "enabled",
 					budget_tokens: options.thinking.budgetTokens || 1024,
@@ -194,14 +200,16 @@ export class AnthropicLLM implements LLM<AnthropicLLMOptions> {
 				thinking,
 				thinkingSignature,
 				toolCalls,
-				model: this.model,
+				provider: this.modelInfo.provider,
+				model: this.modelInfo.id,
 				usage,
 				stopReason: this.mapStopReason(msg.stop_reason),
 			};
 		} catch (error) {
 			return {
 				role: "assistant",
-				model: this.model,
+				provider: this.modelInfo.provider,
+				model: this.modelInfo.id,
 				usage: {
 					input: 0,
 					output: 0,

packages/ai/src/providers/google.ts

@@ -11,6 +11,7 @@ import type {
 	LLM,
 	LLMOptions,
 	Message,
+	Model,
 	StopReason,
 	TokenUsage,
 	Tool,
@@ -27,9 +28,9 @@ export interface GoogleLLMOptions extends LLMOptions {
 
 export class GoogleLLM implements LLM<GoogleLLMOptions> {
 	private client: GoogleGenAI;
-	private model: string;
+	private model: Model;
 
-	constructor(model: string, apiKey?: string) {
+	constructor(model: Model, apiKey?: string) {
 		if (!apiKey) {
 			if (!process.env.GEMINI_API_KEY) {
 				throw new Error(
@@ -42,6 +43,10 @@ export class GoogleLLM implements LLM<GoogleLLMOptions> {
 		this.model = model;
 	}
 
+	getModel(): Model {
+		return this.model;
+	}
+
 	async complete(context: Context, options?: GoogleLLMOptions): Promise<AssistantMessage> {
 		try {
 			const contents = this.convertMessages(context.messages);
@@ -71,8 +76,8 @@ export class GoogleLLM implements LLM<GoogleLLMOptions> {
 				};
 			}
 
-			// Add thinking config if enabled
-			if (options?.thinking?.enabled) {
+			// Add thinking config if enabled and model supports it
+			if (options?.thinking?.enabled && this.model.reasoning) {
 				config.thinkingConfig = {
 					includeThoughts: true,
 					...(options.thinking.budgetTokens !== undefined && { thinkingBudget: options.thinking.budgetTokens }),
@@ -81,7 +86,7 @@ export class GoogleLLM implements LLM<GoogleLLMOptions> {
 
 			// Build the request parameters
 			const params: GenerateContentParameters = {
-				model: this.model,
+				model: this.model.id,
 				contents,
 				config,
 			};
@@ -207,14 +212,16 @@ export class GoogleLLM implements LLM<GoogleLLMOptions> {
 				thinking: thinking || undefined,
 				thinkingSignature: thoughtSignature,
 				toolCalls: toolCalls.length > 0 ? toolCalls : undefined,
-				model: this.model,
+				provider: this.model.provider,
+				model: this.model.id,
 				usage,
 				stopReason,
 			};
 		} catch (error) {
 			return {
 				role: "assistant",
-				model: this.model,
+				provider: this.model.provider,
+				model: this.model.id,
 				usage: {
 					input: 0,
 					output: 0,

packages/ai/src/providers/openai-completions.ts

@@ -6,6 +6,7 @@ import type {
 	LLM,
 	LLMOptions,
 	Message,
+	Model,
 	StopReason,
 	TokenUsage,
 	Tool,
@@ -19,9 +20,9 @@ export interface OpenAICompletionsLLMOptions extends LLMOptions {
 
 export class OpenAICompletionsLLM implements LLM<OpenAICompletionsLLMOptions> {
 	private client: OpenAI;
-	private model: string;
+	private modelInfo: Model;
 
-	constructor(model: string, apiKey?: string, baseUrl?: string) {
+	constructor(model: Model, apiKey?: string) {
 		if (!apiKey) {
 			if (!process.env.OPENAI_API_KEY) {
 				throw new Error(
@@ -30,8 +31,12 @@ export class OpenAICompletionsLLM implements LLM<OpenAICompletionsLLMOptions> {
 			}
 			apiKey = process.env.OPENAI_API_KEY;
 		}
-		this.client = new OpenAI({ apiKey, baseURL: baseUrl });
-		this.model = model;
+		this.client = new OpenAI({ apiKey, baseURL: model.baseUrl });
+		this.modelInfo = model;
+	}
+
+	getModel(): Model {
+		return this.modelInfo;
 	}
 
 	async complete(request: Context, options?: OpenAICompletionsLLMOptions): Promise<AssistantMessage> {
@@ -39,14 +44,14 @@ export class OpenAICompletionsLLM implements LLM<OpenAICompletionsLLMOptions> {
 			const messages = this.convertMessages(request.messages, request.systemPrompt);
 
 			const params: OpenAI.Chat.Completions.ChatCompletionCreateParamsStreaming = {
-				model: this.model,
+				model: this.modelInfo.id,
 				messages,
 				stream: true,
 				stream_options: { include_usage: true },
 			};
 
 			// Cerebras/xAI dont like the "store" field
-			if (!this.client.baseURL?.includes("cerebras.ai") || this.client.baseURL?.includes("api.x.ai")) {
+			if (!this.modelInfo.baseUrl?.includes("cerebras.ai") && !this.modelInfo.baseUrl?.includes("api.x.ai")) {
 				(params as any).store = false;
 			}
 
@@ -66,7 +71,11 @@ export class OpenAICompletionsLLM implements LLM<OpenAICompletionsLLMOptions> {
 				params.tool_choice = options.toolChoice;
 			}
 
-			if (options?.reasoningEffort && this.isReasoningModel() && !this.model.toLowerCase().includes("grok")) {
+			if (
+				options?.reasoningEffort &&
+				this.modelInfo.reasoning &&
+				!this.modelInfo.id.toLowerCase().includes("grok")
+			) {
 				params.reasoning_effort = options.reasoningEffort;
 			}
 
@@ -203,14 +212,16 @@ export class OpenAICompletionsLLM implements LLM<OpenAICompletionsLLMOptions> {
 				thinking: reasoningContent || undefined,
 				thinkingSignature: reasoningField || undefined,
 				toolCalls: toolCalls.length > 0 ? toolCalls : undefined,
-				model: this.model,
+				provider: this.modelInfo.provider,
+				model: this.modelInfo.id,
 				usage,
 				stopReason: this.mapStopReason(finishReason),
 			};
 		} catch (error) {
 			return {
 				role: "assistant",
-				model: this.model,
+				provider: this.modelInfo.provider,
+				model: this.modelInfo.id,
 				usage: {
 					input: 0,
 					output: 0,
@@ -230,9 +241,9 @@ export class OpenAICompletionsLLM implements LLM<OpenAICompletionsLLMOptions> {
 		if (systemPrompt) {
 			// Cerebras/xAi don't like the "developer" role
 			const useDeveloperRole =
-				this.isReasoningModel() &&
-				!this.client.baseURL?.includes("cerebras.ai") &&
-				!this.client.baseURL?.includes("api.x.ai");
+				this.modelInfo.reasoning &&
+				!this.modelInfo.baseUrl?.includes("cerebras.ai") &&
+				!this.modelInfo.baseUrl?.includes("api.x.ai");
 			const role = useDeveloperRole ? "developer" : "system";
 			params.push({ role: role, content: systemPrompt });
 		}
@@ -305,9 +316,4 @@ export class OpenAICompletionsLLM implements LLM<OpenAICompletionsLLMOptions> {
 				return "stop";
 		}
 	}
-
-	private isReasoningModel(): boolean {
-		// TODO base on models.dev
-		return true;
-	}
 }

packages/ai/src/providers/openai-responses.ts

@@ -11,6 +11,7 @@ import type {
 	LLM,
 	LLMOptions,
 	Message,
+	Model,
 	StopReason,
 	TokenUsage,
 	Tool,
@@ -24,9 +25,9 @@ export interface OpenAIResponsesLLMOptions extends LLMOptions {
 
 export class OpenAIResponsesLLM implements LLM<OpenAIResponsesLLMOptions> {
 	private client: OpenAI;
-	private model: string;
+	private modelInfo: Model;
 
-	constructor(model: string, apiKey?: string, baseUrl?: string) {
+	constructor(model: Model, apiKey?: string) {
 		if (!apiKey) {
 			if (!process.env.OPENAI_API_KEY) {
 				throw new Error(
@@ -35,8 +36,12 @@ export class OpenAIResponsesLLM implements LLM<OpenAIResponsesLLMOptions> {
 			}
 			apiKey = process.env.OPENAI_API_KEY;
 		}
-		this.client = new OpenAI({ apiKey, baseURL: baseUrl });
-		this.model = model;
+		this.client = new OpenAI({ apiKey, baseURL: model.baseUrl });
+		this.modelInfo = model;
+	}
+
+	getModel(): Model {
+		return this.modelInfo;
 	}
 
 	async complete(request: Context, options?: OpenAIResponsesLLMOptions): Promise<AssistantMessage> {
@@ -44,7 +49,7 @@ export class OpenAIResponsesLLM implements LLM<OpenAIResponsesLLMOptions> {
 			const input = this.convertToInput(request.messages, request.systemPrompt);
 
 			const params: ResponseCreateParamsStreaming = {
-				model: this.model,
+				model: this.modelInfo.id,
 				input,
 				stream: true,
 			};
@@ -62,7 +67,7 @@ export class OpenAIResponsesLLM implements LLM<OpenAIResponsesLLMOptions> {
 			}
 
 			// Add reasoning options for models that support it
-			if (this.supportsReasoning() && (options?.reasoningEffort || options?.reasoningSummary)) {
+			if (this.modelInfo?.reasoning && (options?.reasoningEffort || options?.reasoningSummary)) {
 				params.reasoning = {
 					effort: options?.reasoningEffort || "medium",
 					summary: options?.reasoningSummary || "auto",
@@ -145,7 +150,8 @@ export class OpenAIResponsesLLM implements LLM<OpenAIResponsesLLMOptions> {
 				else if (event.type === "error") {
 					return {
 						role: "assistant",
-						model: this.model,
+						provider: this.modelInfo.provider,
+						model: this.modelInfo.id,
 						usage,
 						stopReason: "error",
 						error: `Code ${event.code}: ${event.message}` || "Unknown error",
@@ -159,14 +165,16 @@ export class OpenAIResponsesLLM implements LLM<OpenAIResponsesLLMOptions> {
 				thinking: thinking || undefined,
 				thinkingSignature: JSON.stringify(reasoningItems) || undefined,
 				toolCalls: toolCalls.length > 0 ? toolCalls : undefined,
-				model: this.model,
+				provider: this.modelInfo.provider,
+				model: this.modelInfo.id,
 				usage,
 				stopReason,
 			};
 		} catch (error) {
 			return {
 				role: "assistant",
-				model: this.model,
+				provider: this.modelInfo.provider,
+				model: this.modelInfo.id,
 				usage: {
 					input: 0,
 					output: 0,
@@ -184,7 +192,7 @@ export class OpenAIResponsesLLM implements LLM<OpenAIResponsesLLMOptions> {
 
 		// Add system prompt if provided
 		if (systemPrompt) {
-			const role = this.supportsReasoning() ? "developer" : "system";
+			const role = this.modelInfo?.reasoning ? "developer" : "system";
 			input.push({
 				role,
 				content: systemPrompt,
@@ -260,14 +268,4 @@ export class OpenAIResponsesLLM implements LLM<OpenAIResponsesLLMOptions> {
 				return "stop";
 		}
 	}
-
-	private supportsReasoning(): boolean {
-		// TODO base on models.dev
-		return (
-			this.model.includes("o1") ||
-			this.model.includes("o3") ||
-			this.model.includes("gpt-5") ||
-			this.model.includes("gpt-4o")
-		);
-	}
 }