refactor: remove deprecated max token settings from OpenAIProvider and OpenAIResponseProvider

src/renderer/src/providers/AiProvider/OpenAIProvider.ts

@@ -1139,8 +1139,6 @@ export default class OpenAIProvider extends BaseOpenAIProvider {
     const body = {
       model: model.id,
       messages: [{ role: 'user', content: 'hi' }],
-      max_completion_tokens: 1, // openAI
-      max_tokens: 1, // openAI deprecated 但大部分OpenAI兼容的提供商继续用这个头
       enable_thinking: false, // qwen3
       stream
     }

src/renderer/src/providers/AiProvider/OpenAIResponseProvider.ts

@@ -907,24 +907,18 @@ export abstract class BaseOpenAIProvider extends BaseProvider {
       const response = await this.sdk.responses.create({
         model: model.id,
         input: [{ role: 'user', content: 'hi' }],
-        max_output_tokens: 1,
         stream: true
       })
-      let hasContent = false
       for await (const chunk of response) {
         if (chunk.type === 'response.output_text.delta') {
-          hasContent = true
+          return { valid: true, error: null }
         }
       }
-      if (hasContent) {
-        return { valid: true, error: null }
-      }
       throw new Error('Empty streaming response')
     } else {
       const response = await this.sdk.responses.create({
         model: model.id,
         input: [{ role: 'user', content: 'hi' }],
-        max_output_tokens: 1,
         stream: false
       })
       if (!response.output_text) {