webscout 8.3.1__py3-none-any.whl → 8.3.3__py3-none-any.whl

webscout/Provider/OPENAI/xenai.py

@@ -0,0 +1,514 @@
+import time
+import uuid
+import json
+import random
+import string
+from typing import List, Dict, Optional, Union, Generator, Any
+import warnings
+
+# Use curl_cffi for requests
+from curl_cffi.requests import Session
+from curl_cffi import CurlError
+import urllib3
+
+# Import base classes and utility structures
+from webscout.Provider.OPENAI.base import OpenAICompatibleProvider, BaseChat, BaseCompletions
+from webscout.Provider.OPENAI.utils import (
+    ChatCompletionChunk, ChatCompletion, Choice, ChoiceDelta,
+    ChatCompletionMessage, CompletionUsage
+)
+warnings.filterwarnings("ignore", category=urllib3.exceptions.InsecureRequestWarning)
+# Import LitAgent for user agent generation
+try:
+    from webscout.litagent import LitAgent
+except ImportError:
+    # Define a dummy LitAgent if webscout is not installed or accessible
+    class LitAgent:
+        def random(self) -> str:
+            # Return a default user agent if LitAgent is unavailable
+            print("Warning: LitAgent not found. Using default minimal headers.")
+            return "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/127.0.0.0 Safari/537.36"
+
+# ANSI escape codes for formatting
+BOLD = "\033[1m"
+RED = "\033[91m"
+RESET = "\033[0m"
+
+class Completions(BaseCompletions):
+    def __init__(self, client: 'XenAI'):
+        self._client = client
+
+    def create(
+        self,
+        *,
+        model: str,
+        messages: List[Dict[str, str]],
+        max_tokens: Optional[int] = None,
+        stream: bool = False,
+        temperature: Optional[float] = None,
+        top_p: Optional[float] = None,
+        timeout: Optional[int] = None,
+        proxies: Optional[Dict[str, str]] = None,
+        **kwargs: Any
+    ) -> Union[ChatCompletion, Generator[ChatCompletionChunk, None, None]]:
+        """
+        Creates a model response for the given chat conversation using XenAI API.
+        Mimics openai.chat.completions.create
+        """
+        if model not in self._client.AVAILABLE_MODELS:
+            raise ValueError(f"Model '{model}' not supported by XenAI. Available: {self._client.AVAILABLE_MODELS}")
+
+        # Construct the XenAI-specific payload
+        payload = {
+            "stream": stream,
+            "model": model,
+            "messages": messages,
+            "params": kwargs.get("params", {}),
+            "tool_servers": kwargs.get("tool_servers", []),
+            "features": kwargs.get("features", {"web_search": False}),
+            "chat_id": kwargs.get("chat_id", str(uuid.uuid4())),
+            "id": str(uuid.uuid4()), # Message ID
+            "stream_options": kwargs.get("stream_options", {"include_usage": True})
+        }
+
+        # Add optional OpenAI params to XenAI's 'params' field if provided
+        if temperature is not None: payload["params"]["temperature"] = temperature
+        if top_p is not None: payload["params"]["top_p"] = top_p
+        if max_tokens is not None: payload["params"]["max_tokens"] = max_tokens
+
+        # Generate standard OpenAI-compatible IDs
+        request_id = f"chatcmpl-{uuid.uuid4()}"
+        created_time = int(time.time())
+
+        if stream:
+            return self._create_stream(request_id, created_time, model, payload, timeout, proxies)
+        else:
+            return self._create_non_stream_from_stream(request_id, created_time, model, payload, timeout, proxies)
+
+    def _create_stream(
+        self, request_id: str, created_time: int, model: str, payload: Dict[str, Any], timeout: Optional[int] = None, proxies: Optional[Dict[str, str]] = None
+    ) -> Generator[ChatCompletionChunk, None, None]:
+        """Handles the streaming response from XenAI."""
+        final_usage_data = None # To store usage if received
+        try:
+            response = self._client.session.post(
+                self._client.api_endpoint,
+                headers=self._client.headers,
+                json=payload,
+                stream=True,
+                timeout=timeout or self._client.timeout,
+                proxies=proxies or getattr(self._client, "proxies", None),
+                impersonate="chrome110",
+                verify=False
+            )
+
+            if not response.ok:
+                try:
+                    error_text = response.text
+                except Exception:
+                    error_text = "<Failed to read error response>"
+                raise IOError(
+                    f"XenAI API Error: {response.status_code} {response.reason} - {error_text}"
+                 )
+
+            for line_bytes in response.iter_lines():
+                if line_bytes:
+                    try:
+                        line = line_bytes.decode('utf-8').strip()
+                        if line.startswith("data: "):
+                            json_str = line[6:]
+                            if json_str == "[DONE]":
+                                break # End of stream signal
+
+                            json_data = json.loads(json_str)
+
+                            # Check for usage data in the chunk (often comes near the end)
+                            if 'usage' in json_data and json_data['usage']:
+                                final_usage_data = json_data['usage']
+                                # Don't yield a chunk just for usage, wait for content or final chunk
+
+                            if 'choices' in json_data and len(json_data['choices']) > 0:
+                                choice_data = json_data['choices'][0]
+                                delta_data = choice_data.get('delta', {})
+                                finish_reason = choice_data.get('finish_reason')
+                                content = delta_data.get('content')
+                                role = delta_data.get('role', 'assistant') # Default role
+
+                                # Only yield chunks with content or finish reason
+                                if content is not None or finish_reason:
+                                    delta = ChoiceDelta(content=content, role=role)
+                                    choice = Choice(index=0, delta=delta, finish_reason=finish_reason)
+                                    chunk = ChatCompletionChunk(
+                                        id=request_id,
+                                        choices=[choice],
+                                        created=created_time,
+                                        model=model,
+                                        system_fingerprint=json_data.get('system_fingerprint')
+                                    )
+                                    yield chunk
+
+                    except (json.JSONDecodeError, UnicodeDecodeError):
+                        print(f"{RED}Warning: Could not decode JSON line: {line}{RESET}")
+                        continue
+                    except Exception as e:
+                         print(f"{RED}Error processing stream line: {e} - Line: {line}{RESET}")
+                         continue
+
+            # Final chunk to ensure stream termination is signaled correctly
+            # (even if [DONE] was received, this confirms the generator end)
+            final_delta = ChoiceDelta()
+            # Include usage in the final chunk if available
+            usage_obj = None
+            if final_usage_data:
+                 usage_obj = CompletionUsage(
+                     prompt_tokens=final_usage_data.get('prompt_tokens', 0),
+                     completion_tokens=final_usage_data.get('completion_tokens', 0),
+                     total_tokens=final_usage_data.get('total_tokens', 0),
+                 )
+
+            final_choice = Choice(index=0, delta=final_delta, finish_reason="stop")
+            final_chunk = ChatCompletionChunk(
+                id=request_id,
+                choices=[final_choice],
+                created=created_time,
+                model=model,
+                # system_fingerprint=..., # Can be added if available in final event
+            )
+            # Add usage to the final chunk dictionary representation if available
+            if hasattr(final_chunk, "model_dump"):
+                final_chunk_dict = final_chunk.model_dump(exclude_none=True)
+            else:
+                final_chunk_dict = final_chunk.dict(exclude_none=True)
+            if usage_obj:
+                if hasattr(usage_obj, "model_dump"):
+                    final_chunk_dict["usage"] = usage_obj.model_dump(exclude_none=True)
+                else:
+                    final_chunk_dict["usage"] = usage_obj.dict(exclude_none=True)
+
+            # Yield the final dictionary or object as needed by downstream consumers
+            # Yielding the object aligns better with the generator type hint
+            yield final_chunk
+
+
+        except CurlError as e:
+            print(f"{RED}CurlError during XenAI stream request: {e}{RESET}")
+            raise IOError(f"XenAI request failed due to network/curl issue: {e}") from e
+        except Exception as e:
+            print(f"{RED}Unexpected error during XenAI stream: {e}{RESET}")
+            error_details = ""
+            if hasattr(e, 'response') and e.response is not None:
+                error_details = f" - Status: {e.response.status_code}, Response: {e.response.text}"
+            raise IOError(f"XenAI stream processing failed: {e}{error_details}") from e
+
+    def _create_non_stream_from_stream(
+        self, request_id: str, created_time: int, model: str, payload: Dict[str, Any], timeout: Optional[int] = None, proxies: Optional[Dict[str, str]] = None
+    ) -> ChatCompletion:
+        """Handles the non-streaming response by making a single POST request (like deepinfra)."""
+        try:
+            # Ensure stream is False for non-streaming
+            payload = dict(payload)
+            payload["stream"] = False
+
+            response = self._client.session.post(
+                self._client.api_endpoint,
+                headers=self._client.headers,
+                json=payload,
+                timeout=timeout or self._client.timeout,
+                proxies=proxies or getattr(self._client, "proxies", None),
+                impersonate="chrome110",
+                verify=False
+            )
+            if not response.ok:
+                try:
+                    error_text = response.text
+                except Exception:
+                    error_text = "<Failed to read error response>"
+                raise IOError(
+                    f"XenAI API Error: {response.status_code} {response.reason} - {error_text}"
+                )
+
+            data = response.json()
+            choices_data = data.get('choices', [])
+            usage_data = data.get('usage', {})
+
+            choices = []
+            for choice_d in choices_data:
+                message_d = choice_d.get('message', {})
+                message = ChatCompletionMessage(
+                    role=message_d.get('role', 'assistant'),
+                    content=message_d.get('content', '')
+                )
+                choice = Choice(
+                    index=choice_d.get('index', 0),
+                    message=message,
+                    finish_reason=choice_d.get('finish_reason', 'stop')
+                )
+                choices.append(choice)
+
+            usage = CompletionUsage(
+                prompt_tokens=usage_data.get('prompt_tokens', 0),
+                completion_tokens=usage_data.get('completion_tokens', 0),
+                total_tokens=usage_data.get('total_tokens', 0)
+            )
+
+            completion = ChatCompletion(
+                id=request_id,
+                choices=choices,
+                created=created_time,
+                model=data.get('model', model),
+                usage=usage,
+            )
+            return completion
+
+        except CurlError as e:
+            print(f"{RED}CurlError during XenAI non-stream request: {e}{RESET}")
+            raise IOError(f"XenAI request failed due to network/curl issue: {e}") from e
+        except Exception as e:
+            print(f"{RED}Unexpected error during XenAI non-stream: {e}{RESET}")
+            error_details = ""
+            if hasattr(e, 'response') and e.response is not None:
+                error_details = f" - Status: {e.response.status_code}, Response: {e.response.text}"
+            raise IOError(f"XenAI non-stream processing failed: {e}{error_details}") from e
+
+
+class Chat(BaseChat):
+    def __init__(self, client: 'XenAI'):
+        self.completions = Completions(client)
+
+
+class XenAI(OpenAICompatibleProvider):
+    """
+    OpenAI-compatible client for the XenAI API (chat.XenAI.xyz).
+
+    Usage:
+        client = XenAI()
+        response = client.chat.completions.create(
+            model="google/gemma-7b-it",
+            messages=[{"role": "user", "content": "Hello!"}]
+        )
+        print(response.choices[0].message.content)
+    """
+
+    AVAILABLE_MODELS = [
+        "gemini-2.5-pro-preview-05-06",
+        "gemini-2.5-flash-preview-05-20",
+        "o4-mini-high",
+        "grok-3-mini-fast-beta",
+        "grok-3-fast-beta",
+        "gpt-4.1",
+        "o3-high",
+        "gpt-4o-search-preview",
+        "gpt-4o",
+        "claude-sonnet-4-20250514",
+        "claude-sonnet-4-20250514-thinking",
+        "deepseek-ai/DeepSeek-V3-0324",
+        "deepseek-ai/DeepSeek-R1-0528",
+        "groq/deepseek-r1-distill-llama-70b",
+        "deepseek-ai/DeepSeek-Prover-V2-671B",
+        "meta-llama/llama-4-maverick-17b-128e-instruct",
+        "meta-llama/llama-4-scout-17b-16e-instruct",
+        "cognitivecomputations/Dolphin3.0-Mistral-24B",
+        "sonar-pro",
+        "gpt-4o-mini",
+        "gemini-2.0-flash-lite-preview-02-05",
+        "claude-3-7-sonnet-20250219",
+        "claude-3-7-sonnet-20250219-thinking",
+        "claude-opus-4-20250514",
+        "claude-opus-4-20250514-thinking",
+        "chutesai/Llama-4-Maverick-17B-128E-Instruct-FP8",
+        "chutesai/Llama-4-Scout-17B-16E-Instruct",
+    ]
+
+    def _auto_fetch_token(self):
+        """Automatically fetch a token from the signup endpoint."""
+        session = Session()
+        def random_string(length=8):
+            return ''.join(random.choices(string.ascii_lowercase, k=length))
+        name = random_string(6)
+        email = f"{random_string(8)}@gmail.com"
+        password = email
+        profile_image_url = "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAGQAAABkCAYAAABw4pVUAAAAAXNSR0IArs4c6QAAAkRJREFUeF7tmDFOw0AUBdcSiIaKM3CKHIQ7UHEISq5AiUTFHYC0XADoTRsJEZFEjhFIaYAim92fjGFS736/zOTZzjavl0d98oMh0CgE4+IriEJYPhQC86EQhdAIwPL4DFEIjAAsjg1RCIwALI4NUQiMACyODVEIjAAsjg1RCIwALI4NUQiMACyODVEIjAAsjg1RCIwALI4NUQiMACyODVEIjAAsjg1RCIwALI4NUQiMACyODVEIjAAsjg2BCfkAIqwAA94KZ/EAAAAASUVORK5CYII="
+        payload = {
+            "name": name,
+            "email": email,
+            "password": password,
+            "profile_image_url": profile_image_url
+        }
+        # Add more detailed browser-like headers
+        try:
+            # First try with fingerprint from LitAgent
+            headers = {
+                **LitAgent().generate_fingerprint(),
+                'origin': 'https://chat.xenai.tech',
+                'referer': 'https://chat.xenai.tech/auth',
+                'sec-ch-ua': '"Google Chrome";v="127", "Chromium";v="127", "Not=A?Brand";v="24"',
+                'sec-ch-ua-mobile': '?0',
+                'sec-ch-ua-platform': '"Windows"',
+                'sec-fetch-dest': 'empty',
+                'sec-fetch-mode': 'cors',
+                'sec-fetch-site': 'same-origin',
+                'accept-language': 'en-US,en;q=0.9'
+            }
+        except:
+            # Fallback to basic Chrome user agent if LitAgent fails
+            headers = {
+                'user-agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/127.0.0.0 Safari/537.36',
+                'origin': 'https://chat.xenai.tech',
+                'referer': 'https://chat.xenai.tech/auth',
+                'sec-ch-ua': '"Google Chrome";v="127", "Chromium";v="127", "Not=A?Brand";v="24"',
+                'sec-ch-ua-mobile': '?0',
+                'sec-ch-ua-platform': '"Windows"',
+                'sec-fetch-dest': 'empty',
+                'sec-fetch-mode': 'cors',
+                'sec-fetch-site': 'same-origin',
+                'accept-language': 'en-US,en;q=0.9'
+            }
+        try:
+            # Try signup with newer Chrome version
+            resp = session.post(
+                "https://chat.xenai.tech/api/v1/auths/signup",
+                headers=headers,
+                json=payload,
+                timeout=30,
+                impersonate="chrome120",  # Try newer Chrome version
+                verify=False
+            )
+            
+            if resp.ok:
+                data = resp.json()
+                token = data.get("token")
+                if token:
+                    return token
+                # fallback: try to get from set-cookie
+                set_cookie = resp.headers.get("set-cookie", "")
+                if "token=" in set_cookie:
+                    return set_cookie.split("token=")[1].split(";")[0]
+            
+            # If signup fails, try login (account might already exist)
+            login_resp = session.post(
+                "https://chat.xenai.tech/api/v1/auths/login",
+                headers=headers,
+                json={"email": email, "password": password},
+                timeout=30,
+                impersonate="chrome120",
+                verify=False
+            )
+            
+            if login_resp.ok:
+                data = login_resp.json()
+                token = data.get("token")
+                if token:
+                    return token
+                
+            # Try guest authentication as last resort
+            guest_resp = session.post(
+                "https://chat.xenai.tech/api/v1/auths/guest",
+                headers=headers,
+                json={},
+                timeout=30,
+                impersonate="chrome120",
+                verify=False
+            )
+            
+            if guest_resp.ok:
+                data = guest_resp.json()
+                token = data.get("token")
+                if token:
+                    return token
+                    
+            raise RuntimeError(f"Failed to auto-fetch token: {resp.status_code} {resp.text}")
+        except Exception as e:
+            raise RuntimeError(f"Token auto-fetch failed: {e}")
+
+    def __init__(
+        self,
+        timeout: int = 60,
+    ):
+        """
+        Initializes the XenAI OpenAI-compatible client.
+
+        Args:
+            timeout: Request timeout in seconds.
+        """
+        self.api_endpoint = "https://chat.xenai.tech/api/chat/completions"
+        self.timeout = timeout
+        self.token = self._auto_fetch_token()
+        self.session = Session() # Use curl_cffi Session
+
+        # Enhanced headers with browser impersonation
+        try:
+            self.headers = {
+                **LitAgent().generate_fingerprint(),
+                'origin': 'https://chat.xenai.tech',
+                'referer': 'https://chat.xenai.tech/',
+                'sec-ch-ua': '"Google Chrome";v="127", "Chromium";v="127", "Not=A?Brand";v="24"',
+                'sec-ch-ua-mobile': '?0',
+                'sec-ch-ua-platform': '"Windows"',
+                'sec-fetch-dest': 'empty',
+                'sec-fetch-mode': 'cors',
+                'sec-fetch-site': 'same-origin',
+                'accept-language': 'en-US,en;q=0.9'
+            }
+        except:
+            self.headers = {
+                'user-agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/127.0.0.0 Safari/537.36',
+                'origin': 'https://chat.xenai.tech',
+                'referer': 'https://chat.xenai.tech/',
+                'sec-ch-ua': '"Google Chrome";v="127", "Chromium";v="127", "Not=A?Brand";v="24"',
+                'sec-ch-ua-mobile': '?0',
+                'sec-ch-ua-platform': '"Windows"',
+                'sec-fetch-dest': 'empty',
+                'sec-fetch-mode': 'cors',
+                'sec-fetch-site': 'same-origin',
+                'accept-language': 'en-US,en;q=0.9'
+            }
+
+        # Update headers and set authorization token
+        self.headers['authorization'] = f'Bearer {self.token}'
+        self.session.headers.update(self.headers)
+        
+        # Configure session
+        self.session.impersonate = "chrome120"
+        self.session.verify = False
+        
+        self.chat = Chat(self) # Initialize chat interface
+
+    @property
+    def models(self):
+        class _ModelList:
+            def list(inner_self):
+                return type(self).AVAILABLE_MODELS
+        return _ModelList()
+
+if __name__ == "__main__":
+    print("-" * 100)
+    print(f"{'Model':<50} {'Status':<10} {'Response'}")
+    print("-" * 100)
+
+    test_prompt = "Say 'Hello' in one word"
+
+    client = XenAI()
+    for model in client.models.list():
+        print(f"\rTesting {model}...", end="")
+        try:
+            presp = client.chat.completions.create(
+                model=model,
+                messages=[{"role": "user", "content": test_prompt}]
+            )
+            # Try to get the response text (truncate to 100 chars)
+            if hasattr(presp, "choices") and presp.choices and hasattr(presp.choices[0], "message"):
+                content = presp.choices[0].message.content or ""
+                clean_text = content.strip().encode('utf-8', errors='ignore').decode('utf-8')
+                display_text = clean_text[:100] + "..." if len(clean_text) > 100 else clean_text
+                status = "✓" if clean_text else "✗"
+                if not clean_text:
+                    display_text = "Empty or invalid response"
+            else:
+                status = "✗"
+                display_text = "Empty or invalid response"
+            print(f"\r{model:<50} {status:<10} {display_text}")
+        except Exception as e:
+            error_msg = str(e)
+            if len(error_msg) > 100:
+                error_msg = error_msg[:97] + "..."
+            print(f"\r{model:<50} {'✗':<10} Error: {error_msg}")

webscout/Provider/OPENAI/yep.py

@@ -5,8 +5,8 @@ import json
 from typing import List, Dict, Optional, Union, Generator, Any
 
 # Import base classes and utility structures
-from .base import OpenAICompatibleProvider, BaseChat, BaseCompletions
-from .utils import (
+from webscout.Provider.OPENAI.base import OpenAICompatibleProvider, BaseChat, BaseCompletions
+from webscout.Provider.OPENAI.utils import (
     ChatCompletionChunk, ChatCompletion, Choice, ChoiceDelta,
     ChatCompletionMessage, CompletionUsage, get_system_prompt, count_tokens  # Import count_tokens
 )
@@ -315,6 +315,12 @@ class YEPCHAT(OpenAICompatibleProvider):
             "Sec-CH-UA-Mobile": "?0",
             "Sec-CH-UA-Platform": f'"{fingerprint["platform"]}"',
             "User-Agent": fingerprint["user_agent"],
+            "x-forwarded-for": fingerprint["x-forwarded-for"],
+            "x-real-ip": fingerprint["x-real-ip"],
+            "x-client-ip": fingerprint["x-client-ip"],
+            "forwarded": fingerprint["forwarded"],
+            "x-forwarded-proto": "https",
+            "x-request-id": fingerprint["x-request-id"],
         }
         self.session.headers.update(self.headers)
 

webscout/Provider/OpenGPT.py

@@ -87,7 +87,7 @@ class OpenGPT(Provider):
     def ask(
         self,
         prompt: str,
-        stream: bool = False, # Note: API does not support streaming
+        stream: bool = False, # Note: API does not support streaming natively
         raw: bool = False,
         optimizer: str = None,
         conversationally: bool = False,
@@ -121,38 +121,54 @@ class OpenGPT(Provider):
             "id": self.app_id,
             "userKey": ""  # Assuming userKey is meant to be empty as in the original code
         }
-        
-        # API does not stream, implement non-stream logic directly
+
+        def for_stream():
+            try:
+                response = self.session.post(
+                    "https://open-gpt.app/api/generate",
+                    data=json.dumps(payload),
+                    timeout=self.timeout,
+                    impersonate="chrome110"
+                )
+                response.raise_for_status()
+                response_text = response.text
+                buffer = ""
+                chunk_size = 32
+                for i in range(0, len(response_text), chunk_size):
+                    out = response_text[i:i+chunk_size]
+                    if out.strip():
+                        if raw:
+                            yield out
+                        else:
+                            yield {"text": out}
+                self.last_response = {"text": response_text}
+                self.conversation.update_chat_history(prompt, response_text)
+            except CurlError as e:
+                raise exceptions.FailedToGenerateResponseError(f"Request failed (CurlError): {e}") from e
+            except Exception as e:
+                err_text = getattr(e, 'response', None) and getattr(e.response, 'text', '')
+                raise exceptions.FailedToGenerateResponseError(f"An unexpected error occurred ({type(e).__name__}): {e} - {err_text}") from e
+
         def for_non_stream():
             try:
-                # Use curl_cffi session post with impersonate
                 response = self.session.post(
                     "https://open-gpt.app/api/generate",
-                    # headers are set on the session
-                    data=json.dumps(payload), # Keep data as JSON string
+                    data=json.dumps(payload),
                     timeout=self.timeout,
-                    # proxies are set on the session
-                    impersonate="chrome110" # Use a common impersonation profile
+                    impersonate="chrome110"
                 )
-                
-                response.raise_for_status() # Check for HTTP errors
-                
-                # Use response.text which is already decoded
+                response.raise_for_status()
                 response_text = response.text
                 self.last_response = {"text": response_text}
                 self.conversation.update_chat_history(prompt, response_text)
-                
-                # Return dict or raw string based on raw flag
                 return {"raw": response_text} if raw else {"text": response_text}
-                
-            except CurlError as e: # Catch CurlError
+            except CurlError as e:
                 raise exceptions.FailedToGenerateResponseError(f"Request failed (CurlError): {e}") from e
-            except Exception as e: # Catch other potential exceptions (like HTTPError, JSONDecodeError)
+            except Exception as e:
                 err_text = getattr(e, 'response', None) and getattr(e.response, 'text', '')
                 raise exceptions.FailedToGenerateResponseError(f"An unexpected error occurred ({type(e).__name__}): {e} - {err_text}") from e
-        
-        # This provider doesn't support streaming, so just return non-stream
-        return for_non_stream()
+
+        return for_stream() if stream else for_non_stream()
 
     def chat(
         self,
@@ -173,21 +189,25 @@ class OpenGPT(Provider):
         Returns:
             A string with the response text.
         """
-        # Since ask() now handles both stream=True/False by returning the full response dict:
-        response_data = self.ask(
-            prompt, 
-            stream=False, # Call ask in non-stream mode internally
-            raw=False, # Ensure ask returns dict with 'text' key
-            optimizer=optimizer, 
-            conversationally=conversationally
-        )
-        # If stream=True was requested, simulate streaming by yielding the full message at once
         if stream:
             def stream_wrapper():
-                yield self.get_message(response_data)  # yield only the text string
+                for part in self.ask(
+                    prompt,
+                    stream=True,
+                    raw=False,
+                    optimizer=optimizer,
+                    conversationally=conversationally
+                ):
+                    yield self.get_message(part) if isinstance(part, dict) else part
             return stream_wrapper()
         else:
-            # If stream=False, return the full message directly
+            response_data = self.ask(
+                prompt,
+                stream=False,
+                raw=False,
+                optimizer=optimizer,
+                conversationally=conversationally
+            )
             return self.get_message(response_data)
 
     def get_message(self, response: dict) -> str:
@@ -206,4 +226,6 @@ class OpenGPT(Provider):
 
 if __name__ == "__main__":
     ai = OpenGPT()
-    print(ai.chat("Hello, how are you?"))
+    response = ai.chat("write me about humans in points", stream=True)
+    for part in response:
+        print(part, end="", flush=True)