Create model candidates
01
Foundation Models
Evolving the models themselves — merging fine-tuned expert models into stronger ones without further fine-tuning.
EvoGM · Evolutionary model mergingThe Infrastructure for Evolutionary AI Systems
Enabling intelligence to adapt, create, and evolve across digital and physical worlds.
Latest News
View All NewsGenesis 0.14.1: Keep Model Choices Attached to the Work
Account-level model profiles, subscription authentication and a shared remote server reshape configuration in Genesis 0.14.1. Here is what changes and what to check when upgrading.
Genesis Development: Give Projects a Persistent Registry
November 9 changes on main connect project navigation to a persistent VCT registry. Stable mount identity, host-specific loading and a guarded migration make the local-to-remote handoff more explicit.
Genesis Assistant: A Clearer View of Running Work
Inspect tasks, understand system capacity and act through one conversation. Genesis’s recent main-branch changes also simplify remote services, background jobs and model credentials.
01Why Evolution
The Power of Evolution
Evolution is nature's most proven way to create intelligence. EvoX brings that loop to AI: generate many candidates, test them for real, keep what works, and build on it — so capability compounds through experiment instead of hand-tuning.
Create diverse candidates; in later rounds, inherit and vary retained solutions.
02Applications
Evolution Across AI Paradigms
From models to agents, robots, and scientific discovery — one evolutionary loop, applied wherever intelligence needs to adapt.
02Featured
The root agent branches from main
Agentic AI
Recursive agents that decompose, build, and evolve entire codebases — from a blank repository to shipped software.
Genesis · Recursive agentic system03
Policies control stride, lift, balance and timing
Physical AI
Evolving robot control policies through massively parallel interaction with simulated environments.
EvoRL · Evolutionary reinforcement learning04
Propose airfoils with different camber and thickness
AI for Science
Data-driven black-box optimization for scientific and engineering design, validated on supercritical airfoils.
EvoGO · Supercritical airfoil design03Hardware
Computing Without Boundaries
Turn batches of complex tasks into tensor computation.
01Batched complex tasks
02EvoX · Tensorization
03Heterogeneous computing
CPUGPU / NPU
- NVIDIA GPUSupported
- AMD GPUSupported via ROCm
- VulkanSupported via EvoX Compiler
- Apple SiliconSupported via MPS
- Huawei AscendProject-level validation
- CPUSupported
- DistributedExperimental