@@ -47,9 +47,9 @@ class Chatgpt(Chatgpt_4, StreamSimpleService, OpenAIStreamMixin):
             'input': Decimal('0.0025'),  # $5 / 1M tokens
             'output': Decimal('0.015'),  # $30 / 1M tokens
             'web_search': {
-                'low': Decimal('5'),  # 1 call
-                'medium': Decimal('5'),  # 1 call
-                'high': Decimal('5'),  # 1 call
+                'low': Decimal('5'),  # $0.01 / call
+                'medium': Decimal('5'),  # $0.01 / call
+                'high': Decimal('5'),  # $0.01 / call
             },
             'code_interpreter': Decimal('15'),  # 1 call
             'generated_image': Decimal('10.2'),
@@ -58,31 +58,31 @@ class Chatgpt(Chatgpt_4, StreamSimpleService, OpenAIStreamMixin):
             'input': Decimal('0.0025'),  # $5 / 1M tokens
             'output': Decimal('0.015'),  # $30 / 1M tokens
             'web_search': {
-                'low': Decimal('5'),  # 1 call
-                'medium': Decimal('5'),  # 1 call
-                'high': Decimal('5'),  # 1 call
+                'low': Decimal('5'),  # $0.01 / call
+                'medium': Decimal('5'),  # $0.01 / call
+                'high': Decimal('5'),  # $0.01 / call
             },
             'code_interpreter': Decimal('15'),  # 1 call
             'generated_image': Decimal('10.2'),
         },
         'gpt-5.6-luna': {
-            'input': Decimal('0.0005'),  # $1 / 1M tokens
-            'output': Decimal('0.003'),  # $6 / 1M tokens
+            'input': Decimal('0.0001'),  # $0.2 / 1M tokens
+            'output': Decimal('0.0006'),  # $1.2 / 1M tokens
             'web_search': {
-                'low': Decimal('5'),  # 1 call
-                'medium': Decimal('5'),  # 1 call
-                'high': Decimal('5'),  # 1 call
+                'low': Decimal('5'),  # $0.01 / call
+                'medium': Decimal('5'),  # $0.01 / call
+                'high': Decimal('5'),  # $0.01 / call
             },
             'code_interpreter': Decimal('15'),  # 1 call
             'generated_image': Decimal('10.2'),
         },
         'gpt-5.6-terra': {
-            'input': Decimal('0.00125'),  # $2.5 / 1M tokens
-            'output': Decimal('0.0075'),  # $15 / 1M tokens
+            'input': Decimal('0.001'),  # $2 / 1M tokens
+            'output': Decimal('0.006'),  # $12 / 1M tokens
             'web_search': {
-                'low': Decimal('5'),  # 1 call
-                'medium': Decimal('5'),  # 1 call
-                'high': Decimal('5'),  # 1 call
+                'low': Decimal('5'),  # $0.01 / call
+                'medium': Decimal('5'),  # $0.01 / call
+                'high': Decimal('5'),  # $0.01 / call
             },
             'code_interpreter': Decimal('15'),  # 1 call
             'generated_image': Decimal('10.2'),
@@ -493,7 +493,9 @@ class Chatgpt(Chatgpt_4, StreamSimpleService, OpenAIStreamMixin):
                     Decimal(serper_sources * 250) / Decimal(2.7) * self.TOKENS_COST[model_name]['input']
                 )
         if info.get('code_interpreter'):
-            json_data['tools'].append({'type': 'code_interpreter', 'container': {'type': 'auto'}})
+            json_data['tools'].append(
+                {'type': 'code_interpreter', 'container': {'type': 'auto', 'memory_limit': '1g'}}
+            )
             messages[-1]['content'] += ' the python tool '
             predicted_input_price += self.TOKENS_COST[model_name]['code_interpreter']
         if ctx['image']:
@@ -67,18 +67,18 @@ class Chatgpt_4(SimpleService):
             'input': Decimal('0.0003'),
             'output': Decimal('0.0003'),
             'web_search': {
-                'low': Decimal('12.5'),  # 1 call
-                'medium': Decimal('13.75'),  # 1 call
-                'high': Decimal('15'),  # 1 call
+                'low': Decimal('5'),  # $0.01 / call
+                'medium': Decimal('5'),  # $0.01 / call
+                'high': Decimal('5'),  # $0.01 / call
             },
         },
         'gpt-4o': {
             'input': Decimal('0.005'),
             'output': Decimal('0.005'),
             'web_search': {
-                'low': Decimal('15'),  # 1 call
-                'medium': Decimal('17.5'),  # 1 call
-                'high': Decimal('25'),  # 1 call
+                'low': Decimal('5'),  # $0.01 / call
+                'medium': Decimal('5'),  # $0.01 / call
+                'high': Decimal('5'),  # $0.01 / call
             },
         },
         'gpt-oss-120b': {'input': Decimal('0.0002'), 'output': Decimal('0.0002')},
@@ -566,7 +566,7 @@ class Chatgpt_4(SimpleService):
             'input': messages,
             'tools': [
                 {
-                    'type': 'web_search_preview',
+                    'type': 'web_search',
                     'search_context_size': search_context_size,
                     'user_location': {'type': 'approximate', 'country': 'RU'},
                 }
@@ -26,9 +26,9 @@ class Chatgpt_5(Chatgpt_4):
             'input': Decimal('0.000625'),
             'output': Decimal('0.005'),
             'web_search': {
-                'low': Decimal('5'),  # 1 call
-                'medium': Decimal('5'),  # 1 call
-                'high': Decimal('5'),  # 1 call
+                'low': Decimal('5'),  # $0.01 / call
+                'medium': Decimal('5'),  # $0.01 / call
+                'high': Decimal('5'),  # $0.01 / call
             },
             'code_interpreter': Decimal('15'),
         },
@@ -36,9 +36,9 @@ class Chatgpt_5(Chatgpt_4):
             'input': Decimal('0.000125'),
             'output': Decimal('0.001'),
             'web_search': {
-                'low': Decimal('5'),  # 1 call
-                'medium': Decimal('5'),  # 1 call
-                'high': Decimal('5'),  # 1 call
+                'low': Decimal('5'),  # $0.01 / call
+                'medium': Decimal('5'),  # $0.01 / call
+                'high': Decimal('5'),  # $0.01 / call
             },
             'code_interpreter': Decimal('15'),
         },
@@ -51,9 +51,9 @@ class Chatgpt_5(Chatgpt_4):
             'input': Decimal('0.000625'),
             'output': Decimal('0.005'),
             'web_search': {
-                'low': Decimal('5'),  # 1 call
-                'medium': Decimal('5'),  # 1 call
-                'high': Decimal('5'),  # 1 call
+                'low': Decimal('5'),  # $0.01 / call
+                'medium': Decimal('5'),  # $0.01 / call
+                'high': Decimal('5'),  # $0.01 / call
             },
             'code_interpreter': Decimal('15'),  # 1 call
         },
@@ -61,9 +61,9 @@ class Chatgpt_5(Chatgpt_4):
             'input': Decimal('0.0075'),
             'output': Decimal('0.06'),
             'web_search': {
-                'low': Decimal('5'),  # 1 call
-                'medium': Decimal('5'),  # 1 call
-                'high': Decimal('5'),  # 1 call
+                'low': Decimal('5'),  # $0.01 / call
+                'medium': Decimal('5'),  # $0.01 / call
+                'high': Decimal('5'),  # $0.01 / call
             },
         },
         'gpt-5.1-codex-max': {
@@ -82,9 +82,9 @@ class Chatgpt_5(Chatgpt_4):
             'input': Decimal('0.000875'),
             'output': Decimal('0.007'),
             'web_search': {
-                'low': Decimal('5'),  # 1 call
-                'medium': Decimal('5'),  # 1 call
-                'high': Decimal('5'),  # 1 call
+                'low': Decimal('5'),  # $0.01 / call
+                'medium': Decimal('5'),  # $0.01 / call
+                'high': Decimal('5'),  # $0.01 / call
             },
             'code_interpreter': Decimal('15'),  # 1 call
         },
@@ -239,7 +239,9 @@ class Chatgpt_5(Chatgpt_4):
                 'gpt-5',
                 'gpt-5.1',
             ):
-                json_data['tools'].append({'type': 'code_interpreter', 'container': {'type': 'auto'}})
+                json_data['tools'].append(
+                    {'type': 'code_interpreter', 'container': {'type': 'auto', 'memory_limit': '1g'}}
+                )
                 messages[-1]['content'] += 'the python tool'
             input_tokens, output_tokens, response = self.call_openai_api(
                 proxy=proxy, endpoint='responses', json_data=json_data
@@ -21,9 +21,9 @@ class Chatgpt_5_4(Chatgpt):
             'input': Decimal('0.00125'),
             'output': Decimal('0.0075'),
             'web_search': {
-                'low': Decimal('5'),
-                'medium': Decimal('5'),
-                'high': Decimal('5'),
+                'low': Decimal('5'),  # $0.01 / call
+                'medium': Decimal('5'),  # $0.01 / call
+                'high': Decimal('5'),  # $0.01 / call
             },
             'code_interpreter': Decimal('15'),
             'generated_image': Decimal('10.2'),
@@ -32,9 +32,9 @@ class Chatgpt_5_4(Chatgpt):
             'input': Decimal('0.0075'),
             'output': Decimal('0.045'),
             'web_search': {
-                'low': Decimal('5'),
-                'medium': Decimal('5'),
-                'high': Decimal('5'),
+                'low': Decimal('5'),  # $0.01 / call
+                'medium': Decimal('5'),  # $0.01 / call
+                'high': Decimal('5'),  # $0.01 / call
             },
             'generated_image': Decimal('10.2'),
         },
@@ -227,7 +227,11 @@ class Claude(SerperMixin, StreamSimpleService):
         return memory
 
     def _build_callback_data(self, input_message: Message) -> dict[str, Any]:
-        return {'provider': {'order': ['anthropic']}, **input_message.info, 'tools': []}
+        return {
+            **input_message.info,
+            'provider': {'order': ['anthropic'], 'allow_fallbacks': False},
+            'tools': [],
+        }
 
     def _prepare_messages(
         self, input_message: Message, version_slug: str, callback_data: dict
@@ -26,12 +26,12 @@ class Deepseek(SimpleService):
         #     'output': Decimal('0'),
         # },
         'deepseek/deepseek-v4-pro': {
-            'input': Decimal('1050') / 1_000_000,
-            'output': Decimal('2200') / 1_000_000,
+            'input': Decimal('261') / 1_000_000,  # $0.87 / 1M tokens
+            'output': Decimal('522') / 1_000_000,  # $1.74 / 1M tokens
         },
-        'deepseek/deepseek-v4-flash': {
-            'input': Decimal('100') / 1_000_000,
-            'output': Decimal('175') / 1_000_000,
+        'deepseek/deepseek-v4-flash-0731': {
+            'input': Decimal('24') / 1_000_000,  # $0.08 / 1M tokens
+            'output': Decimal('75.6') / 1_000_000,  # $0.252 / 1M tokens
         },
     }
 
@@ -62,6 +62,7 @@ class Deepseek(SimpleService):
 
         callback_data = {
             **info,
+            'provider': {'order': ['digitalocean'], 'allow_fallbacks': False},
         }
         messages = [
             {'role': 'system', 'content': system_prompt},
@@ -99,8 +99,8 @@ class Gemini(SimpleService):
             raise ModelVersionNotAvailable(version_slug, self.TOKENS_COST)
         version = f'google/{version_slug}'
         callback_data = {
-            'provider': {'order': ['Google AI Studio']},
             **input_message.info,
+            'provider': {'order': ['google-ai-studio'], 'allow_fallbacks': False},
         }
         messages = self.get_chat_history()
         messages.append({'role': 'user', 'content': input_message.content})
@@ -79,8 +79,8 @@ class Gemini_3_1(StreamSimpleService):
             raise ModelVersionNotAvailable(version_slug, self.TOKENS_COST)
         model_slug = f'google/{version_slug}:online'
         callback_data = {
-            'provider': {'order': ['Google AI Studio']},
             **input_message.info,
+            'provider': {'order': ['google-ai-studio'], 'allow_fallbacks': False},
         }
         messages, embedding_tokens = self._prepare_messages(input_message)
 
@@ -103,8 +103,8 @@ class Gemini_3_1(StreamSimpleService):
             raise ModelVersionNotAvailable(version_slug, self.TOKENS_COST)
         model_slug = f'google/{version_slug}:online'
         callback_data = {
-            'provider': {'order': ['Google AI Studio']},
             **input_message.info,
+            'provider': {'order': ['google-ai-studio'], 'allow_fallbacks': False},
         }
         messages, embedding_tokens = self._prepare_messages(input_message)
         input_tokens = output_tokens = 0
@@ -41,8 +41,8 @@ class Gemma(SimpleService):
 
     def make(self, input_message: Message, save: bool = True) -> list[Message]:
         callback_data = {
-            'provider': {'order': ['DeepInfra']},
             **input_message.info,
+            'provider': {'order': ['deepinfra'], 'allow_fallbacks': False},
         }
         messages = self.get_chat_history()
         messages.append({'role': 'user', 'content': input_message.content})
@@ -53,8 +53,8 @@ class Grok_4_1_Fast(SimpleService):
 
     def make(self, input_message: Message, save: bool = True) -> list[Message]:
         callback_data = {
-            'provider': {'order': ['xAI']},
             **input_message.info,
+            'provider': {'order': ['xai'], 'allow_fallbacks': False},
         }
         messages = self.get_chat_history()
         messages.append({'role': 'user', 'content': input_message.content})
@@ -66,7 +66,10 @@ class Llama(SimpleService):
         if version_slug is None or version_slug not in self.TOKENS_COST:
             raise ModelVersionNotAvailable(version_slug, self.TOKENS_COST)
         version = f'meta-llama/{version_slug}'
-        callback_data = {'provider': {'order': ['DeepInfra']}, **input_message.info}
+        callback_data = {
+            **input_message.info,
+            'provider': {'order': ['deepinfra'], 'allow_fallbacks': False},
+        }
         messages = self.get_chat_history()
         messages.append({'role': 'user', 'content': input_message.content})
         image = input_message.file
@@ -17,9 +17,9 @@ from payments.selectors.payment_plan_selector import PaymentPlanSelector
 
 class Ltx(SimpleService):
     TOKENS_COST = {
-        '1080p': Decimal('12'),
-        '2k': Decimal('24'),
-        '4k': Decimal('48'),
+        '1080p': Decimal('18'),  # $0.06 / sec
+        '2k': Decimal('36'),  # $0.12 / sec
+        '4k': Decimal('72'),  # $0.24 / sec
     }
 
     @classmethod
@@ -55,7 +55,10 @@ class Mistral(SimpleService):
 
     def make(self, input_message: Message, save: bool = True) -> list[Message]:
         version = 'mistralai/mistral-small-3.1-24b-instruct'
-        callback_data = {'provider': {'order': ['Parasail']}, **input_message.info}
+        callback_data = {
+            **input_message.info,
+            'provider': {'order': ['parasail'], 'allow_fallbacks': False},
+        }
         messages = self.get_chat_history()
         messages.append({'role': 'user', 'content': input_message.content})
         image = input_message.file
@@ -69,7 +69,10 @@ class Perplexity(SimpleService):
         if version_slug is None or version_slug not in self.TOKENS_COST:
             raise ModelVersionNotAvailable(version_slug, self.TOKENS_COST)
         version = f'perplexity/{version_slug}'
-        callback_data = {'provider': {'order': ['Perplexity']}, **input_message.info}
+        callback_data = {
+            **input_message.info,
+            'provider': {'order': ['perplexity'], 'allow_fallbacks': False},
+        }
         messages = self.get_chat_history()
         messages.append({'role': 'user', 'content': input_message.content})
         try:
@@ -54,7 +54,10 @@ class Qwen(SimpleService):
         if version_slug is None or version_slug not in self.TOKENS_COST:
             raise ModelVersionNotAvailable(version_slug, self.TOKENS_COST)
         version = f'qwen/{version_slug}'
-        callback_data = {'provider': {'order': ['DeepInfra']}, **input_message.info}
+        callback_data = {
+            **input_message.info,
+            'provider': {'order': ['deepinfra'], 'allow_fallbacks': False},
+        }
         messages = self.get_chat_history()
         messages.insert(
             0,
@@ -58,7 +58,10 @@ class Qwen_235B(SimpleService):
         if version_slug is None or version_slug not in self.TOKENS_COST:
             raise ModelVersionNotAvailable(version_slug, self.TOKENS_COST)
         version = f'qwen/{version_slug}'
-        callback_data = {'provider': {'order': ['DeepInfra']}, **input_message.info}
+        callback_data = {
+            **input_message.info,
+            'provider': {'order': ['deepinfra'], 'allow_fallbacks': False},
+        }
         messages = self.get_chat_history()
         messages.insert(
             0,
@@ -26,8 +26,8 @@ class Qwen_3_7(StreamSimpleService):
     COEFFICIENT = Decimal('300.0')
 
     TOKENS_COST = {
-        'qwen3.7-max': {'input': Decimal('750'), 'output': Decimal('2250')},
-        'qwen3.7-plus': {'input': Decimal('120'), 'output': Decimal('480')},
+        'qwen3.7-max': {'input': Decimal('442.5'), 'output': Decimal('1327.5')},  # $1.475 / $4.425
+        'qwen3.7-plus': {'input': Decimal('96'), 'output': Decimal('384')},  # $0.32 / $1.28
     }
 
     MAX_OUTPUT_TOKENS = 30_000
@@ -38,7 +38,10 @@ class Qwen_3_Max_Thinking(SimpleService):
 
     def make(self, input_message: Message, save: bool = True) -> list[Message]:
         start_time = time.time()
-        callback_data = {'provider': {'order': ['alibaba']}, **input_message.info}
+        callback_data = {
+            **input_message.info,
+            'provider': {'order': ['alibaba'], 'allow_fallbacks': False},
+        }
         messages = self.get_chat_history()
         messages.append({'role': 'user', 'content': input_message.content})
         result = openrouter_run('qwen/qwen3-max-thinking:online', messages, callback_data, 'Qwen')
@@ -20,12 +20,12 @@ from ml_model.tasks import bytedance_model_ark_run
 
 class Reve(SimpleService):
     # Reve временно не работает на репликейте. Временно используем сидрим
-    TEMPORARY_PROVIDER_MODEL = 'seedream-5-0-260128'
+    TEMPORARY_PROVIDER_MODEL = 'seedream-5-0-lite-260128'
 
     PRICE = {
-        '2K': Decimal('25'),
-        '3K': Decimal('50'),
-        '4K': Decimal('100'),
+        '2K': Decimal('17.5'),  # $0.035 / image (seedream-5-0-lite-260128)
+        '3K': Decimal('17.5'),  # $0.035 / image
+        '4K': Decimal('17.5'),  # $0.035 / image
     }
     # PRICE = {
     #     'create': Decimal('12.5'),
@@ -22,18 +22,18 @@ from payments.selectors.payment_plan_selector import PaymentPlanSelector
 class Seedream(SimpleService):
     TOKEN_COST = {
         'seedream-boosted': {
-            '2K': Decimal('25'),
-            '3K': Decimal('50'),
-            '4K': Decimal('100'),
+            '2K': Decimal('17.5'),  # $0.035 / image
+            '3K': Decimal('17.5'),  # $0.035 / image
+            '4K': Decimal('17.5'),  # $0.035 / image
         },
         'seedream-4.5': {
-            '2K': Decimal('25'),
-            '4K': Decimal('100'),
+            '2K': Decimal('20'),  # $0.04 / image
+            '4K': Decimal('20'),  # $0.04 / image
         },
     }
 
     VERSION_MAPPING = {
-        'seedream-boosted': 'seedream-5-0-260128',
+        'seedream-boosted': 'seedream-5-0-lite-260128',
         'seedream-4.5': 'seedream-4-5-251128',
     }
 
@@ -3,4 +3,4 @@ from ml_model.services import Minimaxmusic
 
 
 class Suno(Minimaxmusic):
-    TOKENS_COST = Decimal('17.5')
+    TOKENS_COST = Decimal('15')  # $0.03 * 100 * 5