Method handleO3FamilyMessage

src/api/providers/openai.ts:333–433

(
		modelId: string,
		systemPrompt: string,
		messages: Anthropic.Messages.MessageParam[],
		metadata?: ApiHandlerCreateMessageMetadata,
	)

	}

	private async *handleO3FamilyMessage(
		modelId: string,
		systemPrompt: string,
		messages: Anthropic.Messages.MessageParam[],
		metadata?: ApiHandlerCreateMessageMetadata,
	): ApiStream {
		const modelInfo = this.getModel().info
		const methodIsAzureAiInference = this._isAzureAiInference(this.options.openAiBaseUrl)

		if (this.options.openAiStreamingEnabled ?? true) {
			const isGrokXAI = this._isGrokXAI(this.options.openAiBaseUrl)

			const requestOptions: OpenAI.Chat.Completions.ChatCompletionCreateParamsStreaming = {
				model: modelId,
				messages: [
					{
						role: "developer",
						content: `Formatting re-enabled\n${systemPrompt}`,
					},
					...convertToOpenAiMessages(messages),
				],
				stream: true,
				...(isGrokXAI ? {} : { stream_options: { include_usage: true } }),
				reasoning_effort: modelInfo.reasoningEffort as "low" | "medium" | "high" | undefined,
				temperature: undefined,
				// Tools are always present (minimum ALWAYS_AVAILABLE_TOOLS)
				tools: this.convertToolsForOpenAI(metadata?.tools),
				tool_choice: metadata?.tool_choice,
				parallel_tool_calls: metadata?.parallelToolCalls ?? true,
			}

			// O3 family models do not support the deprecated max_tokens parameter
			// but they do support max_completion_tokens (the modern OpenAI parameter)
			// This allows O3 models to limit response length when includeMaxTokens is enabled
			this.addMaxTokensIfNeeded(requestOptions, modelInfo)

			let stream
			try {
				stream = await this.client.chat.completions.create(
					requestOptions,
					methodIsAzureAiInference ? { path: OPENAI_AZURE_AI_INFERENCE_PATH } : {},
				)
			} catch (error) {
				throw handleOpenAIError(error, this.providerName)
			}

			yield* this.handleStreamResponse(stream)
		} else {
			const requestOptions: OpenAI.Chat.Completions.ChatCompletionCreateParamsNonStreaming = {
				model: modelId,
				messages: [
					{
						role: "developer",
						content: `Formatting re-enabled\n${systemPrompt}`,
					},
					...convertToOpenAiMessages(messages),
				],
				reasoning_effort: modelInfo.reasoningEffort as "low" | "medium" | "high" | undefined,