Qwen 7B Playground

Code version: 2026.11.09-20.55-prism-provider-metadata-V42

Read the README file to view the available features.

Read appunti.md to view tested models.

⭐ Enjoying the Space? Leave a like!

Selected model: Qwen/Qwen2-7B-Instruct

Chatbot

Message

PDF OCR warning / Avviso OCR PDF: scanned PDF pages without selectable text need conversion with LightOnOCR-2-1B. Choose whether to convert below; OCR uses Hugging Face ZeroGPU credits.

Attachments

Convert scanned PDFs? / Convertire i PDF scannerizzati?

LightOnOCR-2-1B conversion uses Hugging Face ZeroGPU credits.

No — do not convert scanned PDFs Yes — convert scanned PDFs with LightOnOCR-2-1B

Upload history: restore a JSON history into this browser session.

The file is checked for valid UTF-8 JSON, a messages list, valid roles and safe size limits before it replaces the current chat. It is not saved persistently.

Inference mode

Local ZeroGPU Use my Hugging Face credits - inference provider

Inference provider models

ZeroGPU size

XLarge uses twice the credit quota and may have a longer queue.

Standard — 48 GB VRAM · 1× credit quota XLarge — 96 GB VRAM · 2× credit quota

ZeroGPU Standard: 48 GB VRAM, ZeroGPU credit quota 1×. This is the default size.

ZeroGPU timeout (seconds)

Maximum time reserved for the local request. A higher value can consume more ZeroGPU quota.

30 150

Search the web with Exa

API key Exa (optional)

Model

ZeroGPU attempts to load this model; it reports an error if the model is incompatible or does not fit in memory.

System prompt

Temperatura

0 1.5

Maximum tokens

32 32768

Input tokens

Output tokens

Query time (s)

Output speed (tok/s)

VRAM: will be detected during a ZeroGPU request.

Local ZeroGPU: uses your Hugging Face profile's ZeroGPU credits to run the model entered in the text box locally.

This Space does not save chat history.