Win a $7K Prize Package Building Apps for This Desktop Robot (15 Days Left)
NVIDIA's GTC Golden Ticket Contest wants your best Reachy Mini application—here's how to enter and what to build
Win a $7K Prize Package Building Apps for This Desktop Robot (15 Days Left)
NVIDIA's GTC Golden Ticket Contest wants your best Reachy Mini application—here's how to enter and what to build
If you've been looking for a reason to dive into embodied AI, NVIDIA just gave you one: a $7,000 prize package including a DGX Spark workstation, VIP access to Jensen Huang's keynote, and an exclusive happy hour at NVIDIA headquarters. All you need to do is build the best Reachy Mini community application in the next 15 days.
The Contest
NVIDIA's GTC 2026 Golden Ticket Developer Contest runs from January 27 to February 15, 2026. Winners get:
- DGX Spark workstation (~$4,000) — 1 PFLOP of AI compute, 128GB memory
- GTC 2026 Conference Pass with VIP keynote seating
- Exclusive NVIDIA HQ Happy Hour
- Training Lab Pass at GTC
The catch: you must be able to travel to San Jose for GTC (March 16-19). But if you can make that work, this is one of the most compelling developer contests running right now.
The Platform: Reachy Mini
Reachy Mini is a $299-449 open-source desktop robot from Pollen Robotics (now owned by Hugging Face). It's 11 inches tall, weighs 3 pounds, and features:
- 6 degrees of freedom head movement
- 360° body rotation
- 2 animated antennas for expression
- Camera, 4 microphones, 5W speaker
- Full Python SDK with Hugging Face Hub integration
Jensen Huang demonstrated Reachy Mini alongside the DGX Spark at CES 2026, calling it "your own personal office R2-D2."
The robot is fully open source—hardware CAD files, firmware, SDK, everything. And here's the key detail: you don't need the physical robot to participate. The MuJoCo simulation environment lets you develop and test without hardware.
The Breakthrough: PersonaPlex Voice AI
What makes this moment special is NVIDIA's PersonaPlex, a speech-to-speech AI that achieves 240ms latency. Unlike traditional voice systems that chain speech-to-text, then LLM processing, then text-to-speech, PersonaPlex works directly with audio tokens.
The result? Truly natural conversation. Andi Marafioti from Hugging Face demonstrated PersonaPlex running on Reachy Mini, and it's genuinely uncanny—no edits, real-time response, the robot listening and speaking simultaneously.
But there's a critical limitation: function calling is "very poor." PersonaPlex is optimized for conversation, not action. As developer @gauravdhiman_ai noted: "It can be good for talking apps like chit chat, but for making real voice AI app that can do something, we need reliable function calling."
This limitation is your opportunity. Projects that solve the function calling problem—or cleverly work around it—will stand out.
What to Build
High-Feasibility Ideas (Play to Strengths)
AI Language Tutor: Use PersonaPlex's low latency for pronunciation practice. The camera enables visual feedback. No function calling needed—pure conversation is the use case.
Meeting Companion: Reachy Mini as a conference room participant that nods, reacts expressively, and provides real-time transcription summaries.
Therapeutic Companion: The expressive movements create emotional connection. Conversation is the core feature. Perfect for elderly care or mental health applications.
Medium-Feasibility Ideas (Work Around Limitations)
Smart Home Hub with Personality: Use external APIs (Home Assistant) for actual device control, while Reachy provides the conversational interface and personality layer.
Interactive Storytelling: Kids don't need function calling—they need engaging stories with expressive physical reactions.
High-Impact Ideas (Requires Innovation)
Function-Calling Middleware: Build the bridge between PersonaPlex and tool execution. This could become foundational infrastructure that other developers need.
How to Enter
Option 1: Social Media Submission
- Follow designated NVIDIA team members and partners (Carter Abdallah, Bryan Catanzaro, Sabrina Koumoin, Pollen Robotics, HuggingFace, Ollama)
- Submit your project to them directly
- Each can award one ticket
Option 2: NVIDIA Hackathon Submit through official NVIDIA hackathons during the entry window.
Getting Started Today
# Install the SDK
pip install reachy-mini[full]
# Run simulation (no hardware needed!)
pip install mujoco
python -m reachy_mini.simulation
The simulation gives you full access to the robot's APIs. You can develop your entire application without waiting the 90 days for hardware delivery.
For PersonaPlex integration:
git clone https://github.com/NVIDIA/personaplex
cd personaplex
pip install moshi/.
Note: PersonaPlex requires an NVIDIA GPU. The demo ran on a 4090.
The Bigger Picture
Jensen Huang declared at CES 2026 that "the ChatGPT moment for physical AI is here." Reachy Mini, at $299, is how everyday developers participate in that moment.
The convergence is happening: low-latency voice AI, open-source robotics, powerful local compute, and 1.7 million Hugging Face models all accessible from a desktop robot. The function calling limitation today is tomorrow's solved problem—and whoever solves it in the next 15 days might win a DGX Spark for their trouble.
Deadline: February 15, 2026 Prize Value: ~$7,000 Requirement: In-person GTC attendance (March 16-19, San Jose)
The clock is ticking. Your desktop robot awaits.
Contest details: developer.nvidia.com/gtc-golden-ticket-contest Reachy Mini SDK: github.com/pollen-robotics/reachy_mini
Written by
Global Builders Club
Global Builders Club
If you found this content valuable, consider donating with crypto.
Suggested Donation: $5-15
Donation Wallet:
0xEc8d88...6EBdF8Accepts:
Supported Chains:
Your support helps Global Builders Club continue creating valuable content and events for the community.



