Method completePrompt

src/api/providers/lite-llm.ts:314–345

(prompt: string)

	}

	async completePrompt(prompt: string): Promise<string> {
		const { id: modelId, info } = await this.fetchModel()

		// Check if this is a GPT-5 model that requires max_completion_tokens instead of max_tokens
		const isGPT5Model = this.isGpt5(modelId)

		try {
			const requestOptions: OpenAI.Chat.Completions.ChatCompletionCreateParamsNonStreaming = {
				model: modelId,
				messages: [{ role: "user", content: prompt }],
			}

			if (this.supportsTemperature(modelId)) {
				requestOptions.temperature = this.options.modelTemperature ?? 0
			}

			// GPT-5 models require max_completion_tokens instead of the deprecated max_tokens parameter
			if (isGPT5Model && info.maxTokens) {
				requestOptions.max_completion_tokens = info.maxTokens
			} else if (info.maxTokens) {
				requestOptions.max_tokens = info.maxTokens
			}

			const response = await this.client.chat.completions.create(requestOptions)
			return response.choices[0]?.message.content || ""
		} catch (error) {
			if (error instanceof Error) {
				throw new Error(`LiteLLM completion error: ${error.message}`)
			}
			throw error
		}
	}
}

// LiteLLM usage may include an extra field for Anthropic use cases.