[firebase_ai] Missing HarmCategory

[ddodi1]

I'm using firebase_ai to generate images using Gemini, as in Firebase documentation.
The same instructions are supplied in Google Cloud - Vertex AI - Create prompt - Get code - Flutter:

import 'dart:typed_data';

import 'package:firebase_ai/firebase_ai.dart';

Future<void> generateContent() async {
  final generationConfig = GenerationConfig(
    maxOutputTokens: 8192,
    temperature: 1,
    topP: 0.95,
  );
  final safetySettings = [
    SafetySetting(
      HarmCategory.hateSpeech,
      HarmBlockThreshold./* TODO(developer): OFF is not supported by the SDK yet. */,
    ),
    SafetySetting(
      HarmCategory.dangerousContent,
      HarmBlockThreshold./* TODO(developer): OFF is not supported by the SDK yet. */,
    ),
    SafetySetting(
      HarmCategory.sexuallyExplicit,
      HarmBlockThreshold./* TODO(developer): OFF is not supported by the SDK yet. */,
    ),
    SafetySetting(
      HarmCategory.harassment,
      HarmBlockThreshold./* TODO(developer): OFF is not supported by the SDK yet. */,
    ),
    SafetySetting(
      HarmCategory./* TODO(developer): HARM_CATEGORY_IMAGE_HATE is not supported by the SDK yet. */,
      HarmBlockThreshold./* TODO(developer): OFF is not supported by the SDK yet. */,
    ),
    SafetySetting(
      HarmCategory./* TODO(developer): HARM_CATEGORY_IMAGE_DANGEROUS_CONTENT is not supported by the SDK yet. */,
      HarmBlockThreshold./* TODO(developer): OFF is not supported by the SDK yet. */,
    ),
    SafetySetting(
      HarmCategory./* TODO(developer): HARM_CATEGORY_IMAGE_HARASSMENT is not supported by the SDK yet. */,
      HarmBlockThreshold./* TODO(developer): OFF is not supported by the SDK yet. */,
    ),
    SafetySetting(
      HarmCategory./* TODO(developer): HARM_CATEGORY_IMAGE_SEXUALLY_EXPLICIT is not supported by the SDK yet. */,
      HarmBlockThreshold./* TODO(developer): OFF is not supported by the SDK yet. */,
    ),
  ];
  final ai = await FirebaseAI.vertexAI.instanceFor(location: 'global');
  final model = ai.generativeModel(
    model: 'gemini-2.0-flash-preview-image-generation',
    generationConfig: generationConfig,
    safetySettings: safetySettings,
  );

  // For multi-turn responses, start a chat session.
  final chat = model.startChat();

  await sendMessage(
    chat,
    Content('user', [
      TextPart('Text...'),
    ]),
  );
}

Future<void> sendMessage(ChatSession chat, Content content) async {
  final response = await chat.sendMessage(content);
  print(response.text);
}

Google Cloud code reports that some HarmCategory's lack, that is the issue I'm facing.

My code:

final safetySettings = [
  SafetySetting(HarmCategory.dangerousContent, HarmBlockThreshold.low, HarmBlockMethod.probability),
  SafetySetting(HarmCategory.harassment, HarmBlockThreshold.low, HarmBlockMethod.probability),
  SafetySetting(HarmCategory.hateSpeech, HarmBlockThreshold.low, HarmBlockMethod.probability),
  SafetySetting(HarmCategory.sexuallyExplicit, HarmBlockThreshold.low, HarmBlockMethod.probability),
];
model = FirebaseAI.vertexAI(location: 'global').generativeModel(
  model: 'gemini-2.0-flash-preview-image-generation',
  safetySettings: safetySettings,
  generationConfig: GenerationConfig(
    responseModalities: [ResponseModalities.text, ResponseModalities.image],
    maxOutputTokens: 8192,
    temperature: 1,
    topP: 0.95,
  ),
);
chat = model.startChat();
final response = await chat.sendMessage(Content.text(textController.text));

Every call I do (with neutral texts) I get an error related to one of the lacking HarmCategory:

E/flutter (10868): [ERROR:flutter/runtime/dart_vm_initializer.cc(40)] Unhandled Exception: FormatException: Unhandled HarmCategory format
E/flutter (10868): HARM_CATEGORY_IMAGE_HATE

Every help would be very appreciated, thanks.

[mbilalbenli]

I am facing same issue. Did you solve that?

[ddodi1]

Not yet