Spaces:
Running
MAJOR UPDATES OF 2025-08-15 IS HERE!
Major updates begin today, August 15, 2025. Here is what is new.
- OpenAI Audio TTS
Every AI response can now be heard in audio form. Please note that due to limitations, if the AI response is too long, the TTS output will be cut off and will not play in full.
Available voices include alloy, echo, fable, onyx, nova, and shimmer. The default voice is nova. You can change the voice in Settings, then Audio, by typing it manually.
Note. This feature is available only in Full Spaces mode.
- Built-in Image Generation from OpenWebUI
Instead of using external scripts, image generation is now directly integrated with the built-in OpenWebUI feature. The model used is Flux and the generated image size is 4096x4096.
Note. This feature is available only in Full Spaces mode.
- Return of Limited Usage Models
Limited usage models are now available again. These include OpenAI GPT-5 and OpenAI o4-mini.
- Return of OpenAI GPT-4.1
This model has input and output limits set by the provider. As a result, some features may not work or responses may be cut off, which can cause errors.
- Built-in Web Search
Instead of using external scripts, web search is now directly integrated using the built-in OpenWebUI feature. The provider for this feature is jina.ai.
Note. This feature is available only in Full Spaces mode.
Additional Information
Normal Spaces refers to accessing Spaces OpenWebUI directly through HuggingFace. In this mode, some OpenWebUI features will not be displayed.
Full Spaces refers to accessing the Spaces link directly through a browser. In this mode, all OpenWebUI features will be displayed.
Moved web search provider to duckduckgo.
Thanks
Major updates begin today, August 15, 2025. Here is what is new.
- OpenAI Audio TTS
Every AI response can now be heard in audio form. Please note that due to limitations, if the AI response is too long, the TTS output will be cut off and will not play in full.
Available voices include alloy, echo, fable, onyx, nova, and shimmer. The default voice is nova. You can change the voice in Settings, then Audio, by typing it manually.
Note. This feature is available only in Full Spaces mode.
- Built-in Image Generation from OpenWebUI
Instead of using external scripts, image generation is now directly integrated with the built-in OpenWebUI feature. The model used is Flux and the generated image size is 4096x4096.
Note. This feature is available only in Full Spaces mode.
- Return of Limited Usage Models
Limited usage models are now available again. These include OpenAI GPT-5 and OpenAI o4-mini.
- Return of OpenAI GPT-4.1
This model has input and output limits set by the provider. As a result, some features may not work or responses may be cut off, which can cause errors.
- Built-in Web Search
Instead of using external scripts, web search is now directly integrated using the built-in OpenWebUI feature. The provider for this feature is jina.ai.
Note. This feature is available only in Full Spaces mode.
Additional Information
Normal Spaces refers to accessing Spaces OpenWebUI directly through HuggingFace. In this mode, some OpenWebUI features will not be displayed.
Full Spaces refers to accessing the Spaces link directly through a browser. In this mode, all OpenWebUI features will be displayed.
Unexpectedly, it turns out that long "plain text" output can be handled by TTS.
Okay, only one issue, every Chat-GPT 4 service [4.1 Nano, regular 4.1, Chat GPT-5] has a 'str' object has no attribute 'get' error
Okay, only one issue, every Chat-GPT 4 service [4.1 Nano, regular 4.1, Chat GPT-5] has a 'str' object has no attribute 'get' error
Could you reproduce?
Yeah, something changes on the Pollinations backend, and OpenWebUI not showing it properly.
I'll check my backend too.
Use non-pollinations if you use long content usage. Since some models have limitation of input/output.
Yeah, something changes on the Pollinations backend, and OpenWebUI not showing it properly.
I'll check my backend too.
Use non-pollinations if you use long content usage. Since some models have limitation of input/output.
Ok, thanks but I have a tendency of generating stuff of over 300, 000 tokens for my fiction, I''l wait
Yeah, something changes on the Pollinations backend, and OpenWebUI not showing it properly.
I'll check my backend too.
Use non-pollinations if you use long content usage. Since some models have limitation of input/output.
Ok, thanks but I have a tendency of generating stuff of over 300, 000 tokens for my fiction, I''l wait
Did you use AI overtime?
You can switch to gpt-5 nano.
Or you can use ChatGPT web/app.
Yeah, something changes on the Pollinations backend, and OpenWebUI not showing it properly.
I'll check my backend too.
Use non-pollinations if you use long content usage. Since some models have limitation of input/output.
Ok, thanks but I have a tendency of generating stuff of over 300, 000 tokens for my fiction, I''l wait
Did you use AI overtime?
You can switch to gpt-5 nano.
Or you can use ChatGPT web/app.
How do I use Chatgpt web/app?