update

update

updatE

updatE

update

update

update

update

update

update

update

update

F

update

update

update

update

update

update

update

update

update

update

update

update

Revert to ZeroGPU configuration

Disable @spaces.GPU decorator for T4 dedicated GPU

Remove unsupported Gradio parameters

Add loading indicator and Arabic UI improvements

Hide thinking process by default in chat messages

Revert to default launch settings for maximum compatibility

Fix compatibility issues with Gradio on Spaces

Fix RTL support for Arabic interface

Update page title and favicon for NB TEL branding

Fix system prompt handling after removing UI field

Correct model name in description to anaspro/Lahja-iraqi-4B

Remove System Prompt from Additional Inputs

Switch to Unsloth GPT-OSS-20B 4-bit quantized model

Add NB TEL system prompt integration and Arabic UI updates

F

Use original simple code structure with our customizations

Add model caching with lru_cache for ZeroGPU

Fix ZeroGPU compatibility - load model inside GPU context

Implement official Harmony format with openai-harmony package

update

update

Fix model loading - load on startup not during chat

Update system prompt and interface for GPT-OSS testing

Update to GPT-OSS-20B model with 4-bit quantization

Update app.py