pygpt-net 2.5.17__py3-none-any.whl → 2.5.19__py3-none-any.whl

pygpt_net/provider/gpt/image.py

@@ -6,9 +6,9 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2024.12.14 22:00:00                  #
+# Updated Date: 2025.06.26 18:00:00                  #
 # ================================================== #
-
+import base64
 import datetime
 import os
 from typing import Optional, Dict, Any
@@ -132,6 +132,7 @@ class ImageWorker(QObject, QRunnable):
         self.allowed_max_num = {
             "dall-e-2": 4,
             "dall-e-3": 1,
+            "gpt-image-1": 1,
         }
         self.allowed_resolutions = {
             "dall-e-2": [
@@ -144,6 +145,27 @@ class ImageWorker(QObject, QRunnable):
                 "1024x1792",
                 "1024x1024",
             ],
+            "gpt-image-1": [
+                "1536x1024",
+                "1024x1536",
+                "1024x1024",
+                "auto",
+            ],
+        }
+        self.allowed_quality = {
+            "dall-e-2": [
+                "standard",
+            ],
+            "dall-e-3": [
+                "standard",
+                "hd",
+            ],
+            "gpt-image-1": [
+                "auto",
+                "high",
+                "medium",
+                "low",
+            ],
         }
 
     @Slot()
@@ -188,6 +210,11 @@ class ImageWorker(QObject, QRunnable):
                 if resolution not in self.allowed_resolutions[self.model]:
                     resolution = self.allowed_resolutions[self.model][0]
 
+            quality = self.quality
+            if self.model in self.allowed_quality:
+                if quality not in self.allowed_quality[self.model]:
+                    quality = self.allowed_quality[self.model][0]
+
             # send to API
             response = None
             if self.model == "dall-e-2":
@@ -197,12 +224,12 @@ class ImageWorker(QObject, QRunnable):
                     n=self.num,
                     size=resolution,
                 )
-            elif self.model == "dall-e-3":
+            elif self.model == "dall-e-3" or self.model == "gpt-image-1":
                 response = self.client.images.generate(
                     model=self.model,
                     prompt=self.input_prompt,
                     n=self.num,
-                    quality=self.quality,
+                    quality=quality,
                     size=resolution,
                 )
 
@@ -215,20 +242,27 @@ class ImageWorker(QObject, QRunnable):
             for i in range(self.num):
                 if i >= len(response.data):
                     break
-                url = response.data[i].url
-                res = requests.get(url)
 
                 # generate filename
                 name = datetime.date.today().strftime(
                     "%Y-%m-%d") + "_" + datetime.datetime.now().strftime("%H-%M-%S") + "-" \
-                    + self.window.core.image.make_safe_filename(self.input_prompt) + "-" + str(i + 1) + ".png"
+                       + self.window.core.image.make_safe_filename(self.input_prompt) + "-" + str(i + 1) + ".png"
                 path = os.path.join(self.window.core.config.get_user_dir("img"), name)
 
                 msg = trans('img.status.downloading') + " (" + str(i + 1) + " / " + str(self.num) + ") -> " + str(path)
                 self.signals.status.emit(msg)
 
+                if response.data[i] is None:
+                    self.signals.error.emit("API Error: empty image data")
+                    return
+                if response.data[i].url:  # dall-e 2 and 3 returns URL
+                    res = requests.get(response.data[i].url)
+                    data = res.content
+                else:  # gpt-image-1 returns base64 encoded image
+                    data = base64.b64decode(response.data[i].b64_json)
+
                 # save image
-                if self.window.core.image.save_image(path, res.content):
+                if data and self.window.core.image.save_image(path, data):
                     paths.append(path)
                 else:
                     self.signals.error.emit("Error saving image")

pygpt_net/provider/gpt/responses.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2025.06.25 02:00:00                  #
+# Updated Date: 2025.06.26 18:00:00                  #
 # ================================================== #
 
 import json
@@ -17,13 +17,11 @@ from pygpt_net.core.types import (
     MODE_CHAT,
     MODE_VISION,
     MODE_AUDIO,
-    MODE_RESEARCH,
 )
 from pygpt_net.core.bridge.context import BridgeContext, MultimodalContext
 from pygpt_net.item.ctx import CtxItem
 from pygpt_net.item.model import ModelItem
 
-from .utils import sanitize_name
 from pygpt_net.item.attachment import AttachmentItem
 
 
@@ -38,6 +36,7 @@ class Responses:
         self.input_tokens = 0
         self.audio_prev_id = None
         self.audio_prev_expires_ts = None
+        self.prev_response_id = None
 
     def send(
             self,
@@ -80,6 +79,7 @@ class Responses:
             user_name=user_name,
             multimodal_ctx=multimodal_ctx,
         )
+
         msg_tokens = self.window.core.tokens.from_messages(
             messages,
             model.id,
@@ -116,9 +116,15 @@ class Responses:
             response_kwargs['reasoning']['effort'] = model.extra["reasoning_effort"]
 
         # extend tools with external tools
-        if not model.id.startswith("o1") and not model.id.startswith("o3"):
+        if (not model.id.startswith("o1")
+                and not model.id.startswith("o3")):
             if self.window.core.config.get("remote_tools.web_search", False):
                 tools.append({"type": "web_search_preview"})
+            if self.window.core.config.get("remote_tools.image", False):
+                tool = {"type": "image_generation"}
+                if stream:
+                    tool["partial_images"] = 1  # required for streaming
+                tools.append(tool)
 
         # tool calls are not supported for o1-mini and o1-preview
         if (model.id is not None
@@ -126,18 +132,9 @@ class Responses:
             if len(tools) > 0:
                 response_kwargs['tools'] = tools
 
-        # audio mode
-        if mode in [MODE_AUDIO]:
-            stream = False
-            voice_id = "alloy"
-            tmp_voice = self.window.core.plugins.get_option("audio_output", "openai_voice")
-            if tmp_voice:
-                voice_id = tmp_voice
-            response_kwargs["modalities"] = ["text", "audio"]
-            response_kwargs["audio"] = {
-                "voice": voice_id,
-                "format": "wav"
-            }
+        # attach previous response ID if available
+        if self.prev_response_id:
+            response_kwargs['previous_response_id'] = self.prev_response_id
 
         response = client.responses.create(
             input=messages,
@@ -145,6 +142,11 @@ class Responses:
             stream=stream,
             **response_kwargs,
         )
+
+        # store previous response ID
+        if not stream and response:
+            ctx.msg_id = response.id
+
         return response
 
     def build(
@@ -172,6 +174,7 @@ class Responses:
         :return: messages list
         """
         messages = []
+        self.prev_response_id = None  # reset
 
         # tokens config
         mode = MODE_CHAT
@@ -240,6 +243,9 @@ class Responses:
                                 }
                     messages.append(msg)
 
+                if item.msg_id and (item.cmds is None or len(item.cmds) == 0):  # if no cmds before
+                    self.prev_response_id = item.msg_id  # previous response ID to use in current input
+
         # use vision and audio if available in current model
         content = str(prompt)
         if MODE_VISION in model.mode:

pygpt_net/provider/llms/anthropic.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2024.12.14 22:00:00                  #
+# Updated Date: 2025.06.26 16:00:00                  #
 # ================================================== #
 
 from llama_index.llms.anthropic import Anthropic
@@ -47,4 +47,6 @@ class AnthropicLLM(BaseLLM):
         :return: LLM provider instance
         """
         args = self.parse_args(model.llama_index)
+        if "model" not in args:
+            args["model"] = model.id
         return Anthropic(**args)

pygpt_net/provider/llms/google.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2025.01.16 01:00:00                  #
+# Updated Date: 2025.06.26 16:00:00                  #
 # ================================================== #
 
 from typing import Optional, List, Dict
@@ -51,6 +51,8 @@ class GoogleLLM(BaseLLM):
         :return: LLM provider instance
         """
         args = self.parse_args(model.llama_index)
+        if "model" not in args:
+            args["model"] = model.id
         return Gemini(**args)
 
     def get_embeddings_model(

pygpt_net/provider/llms/hugging_face.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2024.12.14 22:00:00                  #
+# Updated Date: 2025.06.26 16:00:00                  #
 # ================================================== #
 
 from langchain_community.llms import HuggingFaceHub
@@ -39,6 +39,8 @@ class HuggingFaceLLM(BaseLLM):
         :return: LLM provider instance
         """
         args = self.parse_args(model.langchain)
+        if "model" not in args:
+            args["model"] = model.id
         return HuggingFaceHub(**args)
 
     def chat(

pygpt_net/provider/llms/hugging_face_api.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2024.12.14 22:00:00                  #
+# Updated Date: 2025.06.26 16:00:00                  #
 # ================================================== #
 
 import os
@@ -44,6 +44,8 @@ class HuggingFaceApiLLM(BaseLLM):
         :return: LLM provider instance
         """
         args = self.parse_args(model.llama_index)
+        if "model" not in args:
+            args["model"] = model.id
         return HuggingFaceInferenceAPI(**args)
 
     def get_embeddings_model(

pygpt_net/provider/llms/ollama.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2025.06.24 16:00:00                  #
+# Updated Date: 2025.06.26 16:00:00                  #
 # ================================================== #
 
 import os
@@ -66,6 +66,8 @@ class OllamaLLM(BaseLLM):
         :return: LLM provider instance
         """
         args = self.parse_args(model.langchain)
+        if "model" not in args:
+            args["model"] = model.id
         return ChatOllama(**args)
 
     def llama(
@@ -87,7 +89,10 @@ class OllamaLLM(BaseLLM):
         if "request_timeout" not in args:
             args["request_timeout"] = 120
         if 'OLLAMA_API_BASE' in os.environ:
-            args["base_url"] = os.environ['OLLAMA_API_BASE']
+            if "base_url" not in args:
+                args["base_url"] = os.environ['OLLAMA_API_BASE']
+        if "model" not in args:
+            args["model"] = model.id
         return Ollama(**args)
 
     def get_embeddings_model(
@@ -108,7 +113,8 @@ class OllamaLLM(BaseLLM):
                 "args": config,
             })
         if 'OLLAMA_API_BASE' in os.environ:
-            args["base_url"] = os.environ['OLLAMA_API_BASE']
+            if "base_url" not in args:
+                args["base_url"] = os.environ['OLLAMA_API_BASE']
         return OllamaEmbedding(**args)
 
     def init_embeddings(

pygpt_net/provider/llms/openai.py

@@ -6,7 +6,7 @@
 # GitHub:  https://github.com/szczyglis-dev/py-gpt   #
 # MIT License                                        #
 # Created By  : Marcin Szczygliński                  #
-# Updated Date: 2025.01.16 01:00:00                  #
+# Updated Date: 2025.06.26 16:00:00                  #
 # ================================================== #
 
 from typing import Optional, List, Dict
@@ -50,6 +50,8 @@ class OpenAILLM(BaseLLM):
         :return: LLM provider instance
         """
         args = self.parse_args(model.langchain)
+        if "model" not in args:
+            args["model"] = model.id
         return OpenAI(**args)
 
     def chat(
@@ -84,6 +86,8 @@ class OpenAILLM(BaseLLM):
         :return: LLM provider instance
         """
         args = self.parse_args(model.llama_index)
+        if "model" not in args:
+            args["model"] = model.id
         return LlamaOpenAI(**args)
 
     def llama_multimodal(
@@ -101,6 +105,8 @@ class OpenAILLM(BaseLLM):
         :return: LLM provider instance
         """
         args = self.parse_args(model.llama_index)
+        if "model" not in args:
+            args["model"] = model.id
         return LlamaOpenAIMultiModal(**args)
 
     def get_embeddings_model(

pygpt_net/ui/dialog/preset.py

@@ -103,12 +103,12 @@ class Preset(BaseConfigDialog):
         # modes
         mode_keys = [
             MODE_CHAT,
+            MODE_LLAMA_INDEX,
             MODE_AUDIO,
             MODE_COMPLETION,
             MODE_IMAGE,
             MODE_VISION,
             MODE_LANGCHAIN,
-            MODE_LLAMA_INDEX,
             MODE_AGENT_LLAMA,
             MODE_AGENT,
             MODE_EXPERT,

{pygpt_net-2.5.17.dist-info → pygpt_net-2.5.19.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: pygpt-net
-Version: 2.5.17
+Version: 2.5.19
 Summary: Desktop AI Assistant powered by models: OpenAI o1, GPT-4o, GPT-4, GPT-4 Vision, GPT-3.5, DALL-E 3, Llama 3, Mistral, Gemini, Claude, DeepSeek, Bielik, and other models supported by Langchain, Llama Index, and Ollama. Features include chatbot, text completion, image generation, vision analysis, speech-to-text, internet access, file handling, command execution and more.
 License: MIT
 Keywords: py_gpt,py-gpt,pygpt,desktop,app,o1,gpt,gpt4,gpt-4o,gpt-4v,gpt3.5,gpt-4,gpt-4-vision,gpt-3.5,llama3,mistral,gemini,deepseek,bielik,claude,tts,whisper,vision,chatgpt,dall-e,chat,chatbot,assistant,text completion,image generation,ai,api,openai,api key,langchain,llama-index,ollama,presets,ui,qt,pyside
@@ -100,7 +100,7 @@ Description-Content-Type: text/markdown
 
 [![pygpt](https://snapcraft.io/pygpt/badge.svg)](https://snapcraft.io/pygpt)
 
-Release: **2.5.17** | build: **2025-06-25** | Python: **>=3.10, <3.13**
+Release: **2.5.19** | build: **2025-06-27** | Python: **>=3.10, <3.13**
 
 > Official website: https://pygpt.net | Documentation: https://pygpt.readthedocs.io
 > 
@@ -462,9 +462,9 @@ Your API keys will be available here:
 
 **+ Inline Vision and Image generation**
 
-This mode in **PyGPT** mirrors `ChatGPT`, allowing you to chat with models such as `o1`, `GPT-4`, `GPT-4o` and `GPT-3.5`. It works by using the `ChatCompletion` OpenAI API.
+This mode in **PyGPT** mirrors `ChatGPT`, allowing you to chat with models such as `o1`, `o3`, `GPT-4`, `GPT-4o` and `GPT-3.5`. It works by using the `Responses` and `ChatCompletions` OpenAI API. You can select the API to use in: `Config -> Settings -> API Keys -> OpenAI`.
 
-**Tip: This mode directly uses the OpenAI API. If you want to use models other than GPT (such as Gemini, Claude, or Llama3), use `Chat with Files` mode.**
+**Tip: This mode directly uses the OpenAI API. Other models, such as Gemini, Claude, or Llama3, are supported in Chat mode via LlamaIndex, which the application switches to in the background when working with models other than OpenAI.**
 
 The main part of the interface is a chat window where you see your conversations. Below it is a message box for typing. On the right side, you can set up or change the model and system prompt. You can also save these settings as presets to easily switch between models or tasks.
 
@@ -520,7 +520,7 @@ From version `2.0.107` the `davinci` models are deprecated and has been replaced
 
 ### DALL-E 3
 
-**PyGPT** enables quick and easy image creation with `DALL-E 3`. 
+**PyGPT** enables quick and easy image creation with `DALL-E 3` or `gpt-image-1`. 
 The older model version, `DALL-E 2`, is also accessible. Generating images is akin to a chat conversation  -  a user's prompt triggers the generation, followed by downloading, saving to the computer, 
 and displaying the image onscreen. You can send raw prompt to `DALL-E` in `Image generation` mode or ask the model for the best prompt.
 
@@ -1003,7 +1003,7 @@ The name of the currently active profile is shown as (Profile Name) in the windo
 
 ## Built-in models
 
-PyGPT has built-in support for models (as of 2025-06-24):
+PyGPT has built-in support for models (as of 2025-06-27):
 
 - `bielik-11b-v2.3-instruct:Q4_K_M`
 - `chatgpt-4o-latest`
@@ -1049,6 +1049,10 @@ PyGPT has built-in support for models (as of 2025-06-24):
 - `gpt-4o-2024-11-20`
 - `gpt-4o-audio-preview`
 - `gpt-4o-mini`
+- `grok-3`
+- `grok-3-fast`
+- `grok-3-mini`
+- `grok-3-mini-fast`
 - `llama2-uncensored`
 - `llama3.1`
 - `llama3.1:405b`
@@ -1064,6 +1068,8 @@ PyGPT has built-in support for models (as of 2025-06-24):
 - `qwen:7b`
 - `qwen2:7b`
 - `qwen2.5-coder:7b`
+- `qwen3:8b`
+- `qwen3:30b-a3b`
 - `r1` (Perplexity)
 - `sonar` (Perplexity)
 - `sonar-deep-research` (Perplexity)
@@ -1095,7 +1101,7 @@ There is built-in support for those LLM providers:
 
 How to use locally installed Llama 3 or Mistral models:
 
-1) Choose a working mode: `Chat with Files`.
+1) Choose a working mode: `Chat` or `Chat with Files`.
 
 2) On the models list - select, edit, or add a new model (with `ollama` provider). You can edit the model settings through the menu `Config -> Models -> Edit`, then configure the model parameters in the `advanced` section.
 
@@ -4124,6 +4130,18 @@ may consume additional tokens that are not displayed in the main window.
 
 ## Recent changes:
 
+**2.5.19 (2025-06-27)**
+
+- Added option to enable/disable `Responses API` in `Config -> Settings -> API Keys -> OpenAI`.
+- Added support for xAI / Grok models, added grok-3 models.
+
+**2.5.18 (2025-06-26)**
+
+- Non-GPT models are now available in standard Chat mode.
+- Added a new remote tool: `image_generation` in Responses API -> disabled by default, enable in `Config -> Settings -> Remote Tools`. Enables native image generation and editing of uploaded images in Chat mode.
+- Added a new model `gpt-image-1` and improved image generation.
+- Other small fixes.
+
 **2.5.17 (2025-06-25)**
 
 - Added settings for enable/disable Remote Tools via Responses API in Chat mode: Config -> Settings -> Remote tools. Currently only web-search-preview tool is available, rest of tools coming soon.