Performances
RD Popularity Index
RD Growth Index
| 1h | 24h | 7d | 30d | 60d | 90d |
|---|---|---|---|---|---|
| 0.79% | 5.69% | 23.26% | 80.88% | 73.07% | 73.07% |
Comparison
Follow Updates
Follow Lists
People
About flybrain
FlyBrain is an open-source browser-based simulation of the Drosophila melanogaster brain, built on the FlyWire FAFB v783 connectome with 139,255 neurons and 2.7 million connections. It runs a leaky integrate-and-fire model in a Web Worker, allowing real-time interaction: users can place food, touch the fly, blow air, or change lighting, and the fly's behavior emerges from signal propagation through weighted neural connections rather than scripting. The project's core appeal lies in its narrative: a real biological neural network operating in a digital environment, with no pre-programmed behavior. The official website (flybrain.online) hosts the interactive demo, while the X account (@fruitflydev) shares development updates and community experiments.
Its competitive edge stems from using a genuine connectome rather than an artificial neural network, making emergent behaviors—like seeking food or reacting to touch—a direct result of biological wiring. This addresses the market pain point of AI systems that lack interpretability and biological grounding, offering a tangible way to study how neural circuits drive behavior. In the past six months (2026-03-12 to 2026-09-12), the project gained significant traction following Google's release of MaleCNS v1.0 (166,691 neurons, ~125 million synapses) in early September 2026. Developers quickly adapted the connectome for other tasks: DOOMFLY trains a fly brain to play Doom, Stonkfly reads candlestick charts and trades Bitcoin via Coinbase, and others have plugged it into Super Mario 64 or Minecraft. Notably, the original FlyBrain project predates these spin-offs, having been forked from a C. elegans simulator in March 2026.
However, public experiments show limitations: DOOMFLY's latest candidate version has not passed visual, conditioning, or survival tests, and extended training has not produced a stable learning curve—the README explicitly states it has not yet demonstrated survival learning. Similarly, Stonkfly's trading performance remains unverified. These results highlight that while the connectome enables realistic neural dynamics, it does not guarantee adaptive behavior in complex environments.
Positive: a16z co-founder Marc Andreessen followed the project's official X account, drawing market attention and pushing market cap briefly above $55M; the project uses a real fruit fly's neural connectome to train token mechanics, with open-source code and unique narrative; trading fees fund experiments, creating a continuous content loop.
Negative: After the hype, price fell sharply, market cap now ~$12M, down over 60% in 6 hours; liquidity is limited (native pool paired with GOOGL), with concentration risk; project admits neural simulation only handles part of operations, far from independent operation; meme coins lack real use cases, are highly volatile, and most traders lose money.