ComfyUI, NVIDIA, and RunPod Launch $25,000 Open-Source AI Build Challenge
ComfyUI has partnered with NVIDIA, RunPod, MiniMax, LTX, and BFL to launch a 2-week open-source AI build challenge featuring $25,000 in total prizes, including
On October 16, 2026, ComfyUI officially announced an open-source AI build challenge on the Comfy Developer Platform in partnership with NVIDIA AI, RunPod, MiniMax AI, LTX (@ltx_io), and Black Forest Labs (@bfl_ai).
Image source: X @ComfyUI
The initiative puts up $25,000 in prizes—including a $10,000 grand prize and NVIDIA RTX 5090 GPUs—inviting open-source developers worldwide to submit projects over the next two weeks.
$25,000 in Total Prizes and RTX 5090 GPUs: Award Categories
To foster tangible contributions to the open-source generative AI ecosystem, the competition features primary award categories alongside runner-up support:
- Grand Prize: $10,000 awarded to the overall best build demonstrating exceptional technical execution and ecosystem impact
- Most Innovative: An NVIDIA RTX 5090 GPU for the project showcasing the most original architecture or functional extension
- Most Viral: An NVIDIA RTX 5090 GPU for the project achieving outstanding community engagement and reach
- Open Weights Only: An NVIDIA RTX 5090 GPU reserved exclusively for the top project built without proprietary cloud APIs, relying entirely on open-weights models
- Runners-Up: 10 runner-up teams each receive $250 in Comfy Developer Platform credits ($2,500 total)
By establishing a dedicated "Open Weights Only" category, the organizers place a distinct emphasis on fully reproducible, self-hosted open models rather than closed proprietary cloud API wrappers.
Two-Week Submission Window and In-Person SF/LA Events
Submissions opened on October 16 and will run for two weeks through an official Typeform entry portal.
Entrants must submit open-source projects that run natively within the Comfy Developer Platform ecosystem, meeting the technical and evaluation criteria for their respective target categories. Project specifications and community resources are hosted on the challenge landing page (comfy.org/open-source-higgsfield).
Alongside the virtual hackathon, in-person community networking and demo events are scheduled in California:
- San Francisco (SF) Event: October 28, 2026 (Registration via Luma: luma.com/5mydmvu6)
- Los Angeles (LA) Event: October 29, 2026 (Registration via Luma: luma.com/73q6c0ln)
These gatherings provide a venue for participating teams, local developers, and ecosystem partners to exchange project demos and feedback ahead of the submission deadline.
Mobilizing the Open-Source Ecosystem: The Higgsfield Context
The challenge was conceived as a campaign-driven response to commercial AI platform Higgsfield, which had publicly committed to open-sourcing its entire platform but failed to deliver.
In its official announcement, ComfyUI challenged that stance directly: "Higgsfield promised to open-source their entire platform. They didn't. Let's help them." The initiative aims to mobilize community builders and key infrastructure partners to deliver actual open tools rather than unfulfilled promises.
The coalition backing the initiative spans critical compute infrastructure and open model developers, including NVIDIA, RunPod, MiniMax, LTX, and Black Forest Labs. As ComfyUI continues to expand node-based open creation pipelines against closed proprietary services, the challenge serves as a proving ground for the next generation of modular open-source AI builds.
Sources
- ComfyUI Substack: Higgsfield Promised to Open Source
- ComfyUI Official X (@ComfyUI): ComfyUI Open-Source Build Challenge Announcement