BeamVoice Studio
Voice-to-code web studio using Reflection's Beam model.
Speak or type a coding task. Beam-1.0-501B-A23B generates working code you can copy and run.
How Beam is used
The app sends your prompt to Beam-1.0-501B-A23B via Reflection's OpenAI-compatible API (https://api.reflection.ai/openai/v1). Beam's 1M-token context window lets it reason across large codebases.
Run locally
pip install fastapi uvicorn python-dotenv httpx
cp .env.example .env
# add your REFLECTION_API_KEY
uvicorn app.main:app --reload --port 8000
Or open index.html directly in the browser and enter your Reflection API key.
Keys
REFLECTION_API_KEYโ from https://reflection.ai
Optional:
DEEPGRAM_API_KEYโ from https://console.deepgram.com (enables voice recording; the app works without it)
License
MIT
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐ Ask for provider support