| import gradio as gr, os, huggingface_hub as hf | |
| def go(m, q): | |
| c = hf.InferenceClient(api_key=os.environ["HF_TOKEN"]) | |
| s = c.chat.completions.create(model=m, messages=[{"role":"user","content":q}], stream=True) | |
| for i in s: | |
| if i.choices and i.choices[0].delta.content: | |
| yield i.choices[0].delta.content | |
| models = [ | |
| "deepseek-ai/DeepSeek-V3.1", | |
| "meta-llama/Meta-Llama-3-8B-Instruct", | |
| "Qwen/Qwen2.5-7B-Instruct", | |
| "2F-AI/Titan-Atom", | |
| "NobodyExistsOnTheInternet/K3-Q4-GGUF" | |
| ] | |
| gr.Interface( | |
| fn=go, | |
| inputs=[gr.Dropdown(models), gr.Textbox(label="Ask")], | |
| outputs=gr.Textbox(label="Reply"), | |
| title="Multi-Model Chat" | |
| ).launch(share=True) | |